diff --git a/docs/source/As1.rst b/docs/source/As1.rst new file mode 100644 index 000000000..142a8b9da --- /dev/null +++ b/docs/source/As1.rst @@ -0,0 +1,141 @@ +As1 +===== + +This page provides a curated list of functions and properties available in the ``malariagen_data`` API +for data on *Anopheles stephensi* species mosquitoes. + +To set up the API, use the following code:: + + import malariagen_data + as1 = malariagen_data.As1() + +All the functions below can then be accessed as methods on the ``as1`` object. E.g., to call the +``sample_metadata()`` function, do:: + + df_samples = as1.sample_metadata() + +For more information about the data and terms of use, please see the +`MalariaGEN website `_ or contact support@malariagen.net. + +.. currentmodule:: malariagen_data.as1.As1 + +Basic data access +----------------- +.. autosummary:: + :toctree: generated/ + + releases + sample_sets + lookup_release + lookup_study + +Reference genome data access +---------------------------- +.. autosummary:: + :toctree: generated/ + + contigs + genome_sequence + genome_features + plot_transcript + plot_genes + +Sample metadata access +---------------------- +.. autosummary:: + :toctree: generated/ + + sample_metadata + add_extra_metadata + clear_extra_metadata + lookup_sample + count_samples + plot_samples_bar + plot_samples_interactive_map + plot_sample_location_mapbox + plot_sample_location_geo + wgs_data_catalog + cohorts + +SNP data access +--------------- +.. autosummary:: + :toctree: generated/ + + site_mask_ids + snp_calls + snp_allele_counts + plot_snps + site_annotations + is_accessible + biallelic_snp_calls + biallelic_diplotypes + biallelic_snps_to_plink + +SNP frequency analysis +---------------------- +.. autosummary:: + :toctree: generated/ + + snp_allele_frequencies + snp_allele_frequencies_advanced + aa_allele_frequencies + aa_allele_frequencies_advanced + plot_frequencies_heatmap + plot_frequencies_time_series + plot_frequencies_interactive_map + +Principal components analysis (PCA) +----------------------------------- +.. autosummary:: + :toctree: generated/ + + pca + plot_pca_variance + plot_pca_coords + plot_pca_coords_3d + +Genetic distance and neighbour-joining trees (NJT) +-------------------------------------------------- +.. autosummary:: + :toctree: generated/ + + plot_njt + njt + biallelic_diplotype_pairwise_distances + +Heterozygosity analysis +----------------------- +.. autosummary:: + :toctree: generated/ + + plot_heterozygosity + roh_hmm + plot_roh + +Diversity analysis +------------------ +.. autosummary:: + :toctree: generated/ + + cohort_diversity_stats + diversity_stats + plot_diversity_stats + +Diplotype clustering +-------------------- +.. autosummary:: + :toctree: generated/ + + plot_diplotype_clustering + +Fst analysis +------------ +.. autosummary:: + :toctree: generated/ + + average_fst + pairwise_average_fst + plot_pairwise_average_fst + fst_gwss + plot_fst_gwss diff --git a/docs/source/_static/images/anopheles_stephensi.jpg b/docs/source/_static/images/anopheles_stephensi.jpg new file mode 100644 index 000000000..17206cdf1 Binary files /dev/null and b/docs/source/_static/images/anopheles_stephensi.jpg differ diff --git a/docs/source/index.rst b/docs/source/index.rst index 59c7f4110..fd277170a 100644 --- a/docs/source/index.rst +++ b/docs/source/index.rst @@ -33,6 +33,17 @@ API documentation :align: center :width: 100% + .. grid-item-card:: ``As1`` + :link: As1 + :link-type: doc + + *Anopheles stephensi*. + + .. image:: ./_static/images/anopheles_stephensi.jpg + :alt: Anopheles stephensi mosquito mosquito + :align: center + :width: 100% + .. grid-item-card:: ``Amin1`` :link: Amin1 :link-type: doc diff --git a/malariagen_data/__init__.py b/malariagen_data/__init__.py index 48e4eb2d9..69df5dfee 100644 --- a/malariagen_data/__init__.py +++ b/malariagen_data/__init__.py @@ -4,6 +4,7 @@ from .af1 import Af1 from .ag3 import Ag3 from .amin1 import Amin1 +from .as1 import As1 from .anopheles import AnophelesDataResource, Region from .pf7 import Pf7 from .pf8 import Pf8 diff --git a/malariagen_data/as1.py b/malariagen_data/as1.py new file mode 100644 index 000000000..3e1ec56fd --- /dev/null +++ b/malariagen_data/as1.py @@ -0,0 +1,237 @@ +import sys + +import plotly.express as px # type: ignore + +import malariagen_data +from .anopheles import AnophelesDataResource + +MAJOR_VERSION_NUMBER = 1 +MAJOR_VERSION_PATH = "v1.0" +CONFIG_PATH = "v1.0-config.json" +GCS_DEFAULT_URL = "gs://vo_aste_release_master_us_central1/" +GCS_DEFAULT_PUBLIC_URL = "gs://vo_aste_release_master_us_central1/" +GCS_REGION_URLS = { + "us-central1": "gs://vo_aste_release_master_us_central1", +} + +TAXON_PALETTE = px.colors.qualitative.Plotly +TAXON_COLORS = { + "stephensi": TAXON_PALETTE[0], +} + +XPEHH_GWSS_CACHE_NAME = "as1_xpehh_gwss_v1" +IHS_GWSS_CACHE_NAME = "as1_ihs_gwss_v1" +ROH_HMM_CACHE_NAME = "as1_roh_hmm_v1" + + +class As1(AnophelesDataResource): + """Provides access to data from As1.0 releases. + + Parameters + ---------- + url : str, optional + Base path to data. Defaults to use Google Cloud Storage, or can + be a local path on your file system if data have been downloaded. + site_filters_analysis : str, optional + Site filters analysis version. + bokeh_output_notebook : bool, optional + If True (default), configure bokeh to output plots to the notebook. + results_cache : str, optional + Path to directory on local file system to save results. + log : str or stream, optional + File path or stream output for logging messages. + debug : bool, optional + Set to True to enable debug level logging. + show_progress : bool, optional + If True, show a progress bar during longer-running computations. The default can be overridden using an environmental variable named MGEN_SHOW_PROGRESS. + check_location : bool, optional + If True, use ipinfo to check the location of the client system. + **kwargs + Passed through to fsspec when setting up file system access. + + Examples + -------- + Access data from Google Cloud Storage (default): + + >>> import malariagen_data + >>> as1 = malariagen_data.As1() + + Access data downloaded to a local file system: + + >>> as1 = malariagen_data.As1("/local/path/to/vo_as_release/") + + Access data from Google Cloud Storage, with caching on the local file system + in a directory named "gcs_cache": + + >>> as1 = malariagen_data.As1( + ... "simplecache::gs://vo_aste_release_master_us_central1", + ... simplecache=dict(cache_storage="gcs_cache"), + ... ) + + Set up caching of some longer-running computations on the local file system, + in a directory named "results_cache": + + >>> as1 = malariagen_data.As1(results_cache="results_cache") + + """ + + _xpehh_gwss_cache_name = XPEHH_GWSS_CACHE_NAME + _ihs_gwss_cache_name = IHS_GWSS_CACHE_NAME + _roh_hmm_cache_name = ROH_HMM_CACHE_NAME + + def __init__( + self, + url=None, + public_url=GCS_DEFAULT_PUBLIC_URL, + bokeh_output_notebook=True, + results_cache=None, + log=sys.stdout, + debug=False, + show_progress=None, + check_location=True, + cohorts_analysis=None, + site_filters_analysis=None, + discordant_read_calls_analysis=None, + pre=False, + tqdm_class=None, + unrestricted_use_only=False, + surveillance_use_only=False, + **storage_options, + ): + super().__init__( + url=url, + public_url=public_url, + config_path=CONFIG_PATH, + cohorts_analysis=cohorts_analysis, + aim_analysis=None, + aim_metadata_dtype=None, + aim_ids=None, + aim_palettes=None, + site_filters_analysis=site_filters_analysis, + discordant_read_calls_analysis=discordant_read_calls_analysis, + default_site_mask="stephensi", + default_phasing_analysis=None, + default_coverage_calls_analysis=None, + bokeh_output_notebook=bokeh_output_notebook, + results_cache=results_cache, + log=log, + debug=debug, + show_progress=show_progress, + check_location=check_location, + pre=pre, + gcs_default_url=GCS_DEFAULT_URL, + gcs_region_urls=GCS_REGION_URLS, + major_version_number=MAJOR_VERSION_NUMBER, + major_version_path=MAJOR_VERSION_PATH, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + storage_options=storage_options, + tqdm_class=tqdm_class, + taxon_colors=TAXON_COLORS, + virtual_contigs=None, + inversion_tag_path=None, + unrestricted_use_only=unrestricted_use_only, + surveillance_use_only=surveillance_use_only, + ) + + def __repr__(self): + text = ( + f"\n" + f"Storage URL : {self._url}\n" + f"Data releases available : {', '.join(self._available_releases)}\n" + f"Results cache : {self._results_cache}\n" + f"Cohorts analysis : {self._cohorts_analysis}\n" + f"Site filters analysis : {self._site_filters_analysis}\n" + f"Software version : malariagen_data {malariagen_data.__version__}\n" + f"Client location : {self.client_location}\n" + f"Data filtered to unrestricted use only: {self._unrestricted_use_only}\n" + f"Data filtered to surveillance use only: {self._surveillance_use_only}\n" + f"Relevant data releases : {', '.join(self.releases)}\n" + f"---\n" + f"Please note that data are subject to terms of use,\n" + f"for more information see https://www.malariagen.net/data\n" + f"or contact support@malariagen.net. For API documentation see \n" + f"https://malariagen.github.io/malariagen-data-python/v{malariagen_data.__version__}/As1.html" + ) + return text + + def _repr_html_(self): + html = f""" + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
MalariaGEN As1 API client
+ Please note that data are subject to terms of use, + for more information see + the MalariaGEN website or contact support@malariagen.net. + See also the As1 API docs. +
+ Storage URL + {self._url}
+ Data releases available + {", ".join(self._available_releases)}
+ Results cache + {self._results_cache}
+ Cohorts analysis + {self._cohorts_analysis}
+ Site filters analysis + {self._site_filters_analysis}
+ Software version + malariagen_data {malariagen_data.__version__}
+ Client location + {self.client_location}
+ Data filtered for unrestricted use only + {self._unrestricted_use_only}
+ Data filtered for surveillance use only + {self._surveillance_use_only}
+ Relevant data releases + {", ".join(self.releases)}
+ """ + return html diff --git a/tests/anoph/conftest.py b/tests/anoph/conftest.py index 505d17c46..f94768ad3 100644 --- a/tests/anoph/conftest.py +++ b/tests/anoph/conftest.py @@ -3195,6 +3195,286 @@ def init_site_annotations(self): simulate_site_annotations(path=path, genome=self.genome, rng=self.rng) +class As1Simulator(AnophelesSimulator): + def __init__(self, fixture_dir, rng): + super().__init__( + fixture_dir=fixture_dir, + rng=rng, + bucket="vo_aste_release_master_us_central1", + releases=("1.0",), + has_aims=False, + has_cohorts_by_quarter=True, + has_sequence_qc=True, + ) + + def init_config(self): + self.config = { + "PUBLIC_RELEASES": ["1.0"], + "GENESET_GFF3_PATH": "reference/genome/UCISS2018/Anopheles-stephensi-UCISS2018-curated.gff.gz", + "GENOME_FASTA_PATH": "reference/genome/UCISS2018/Anopheles-stephensi-UCISS2018-curated.fasta", + "GENOME_FAI_PATH": "reference/genome/UCISS2018/Anopheles-stephensi-UCISS2018-curated.fasta.fai", + "GENOME_ZARR_PATH": "reference/genome/UCISS2018/Anopheles-stephensi-UCISS2018-curated.zarr", + "GENOME_REF_ID": "UCISS2018", + "GENOME_REF_NAME": "Anopheles stephensi", + "CONTIGS": ["2RL", "3RL", "X"], + "SITE_ANNOTATIONS_ZARR_PATH": "reference/genome/UCISS2018/Anopheles-stephensi-UCISS2018-curated_SEQANNOTATION.zarr", + "DEFAULT_SITE_FILTERS_ANALYSIS": "sc_20260401", + "DEFAULT_COHORTS_ANALYSIS": "20260402", + "DEFAULT_DISCORDANT_READ_CALLS_ANALYSIS": "", + "SITE_MASK_IDS": ["stephensi"], + "PHASING_ANALYSIS_IDS": [], + "COVERAGE_CALLS_ANALYSIS_IDS": ["stephensi"], + } + config_path = self.bucket_path / "v1.0-config.json" + with config_path.open(mode="w") as f: + json.dump(self.config, f, indent=4) + + def init_public_release_manifest(self): + release_path = self.bucket_path / "v1.0" + release_path.mkdir(parents=True, exist_ok=True) + manifest_path = release_path / "manifest.tsv" + manifest = pd.DataFrame( + { + "sample_set": [ + "1365-VO-DJ-ADBI-VMF00318", + "1386-VO-KE-OCHOMO-VMF00339", + "1367-VO-AF-DONNELLY-VMF00320", + ], + "sample_count": [21, 29, 24], + "study_id": [ + "1365-VO-DJ-ADBI", + "1386-VO-KE-OCHOMO", + "1367-VO-AF-DONNELLY", + ], + "study_url": [ + "https://www.malariagen.net/network/where-we-work/1365-VO-DJ-ADBI", + "https://www.malariagen.net/network/where-we-work/1386-VO-KE-OCHOMO", + "https://www.malariagen.net/network/where-we-work/1367-VO-AF-DONNELLY", + ], + "terms_of_use_expiry_date": [ + "2099-12-31", + "2099-12-31", + "2024-01-01", # Set to the past in order to test unrestricted_use_only. + ], + "terms_of_use_url": [ + "https://malariagen.github.io/vector-data/as1/as1.0.html#terms-of-use", + "https://malariagen.github.io/vector-data/as1/as1.0.html#terms-of-use", + "https://malariagen.github.io/vector-data/as1/as1.0.html#terms-of-use", + ], + } + ) + manifest.to_csv(manifest_path, index=False, sep="\t") + self.release_manifests["1.0"] = manifest + + def init_genome_sequence(self): + base_composition = { + b"a": 0.0, + b"c": 0.0, + b"g": 0.0, + b"t": 0.0, + b"n": 0.0, + b"A": 0.29432128333333335, + b"C": 0.20542065, + b"G": 0.20575796666666665, + b"T": 0.2944834333333333, + b"N": 1.6666666666666667e-05, + } + path = self.bucket_path / self.config["GENOME_ZARR_PATH"] + self.genome = simulate_genome( + path=path, + contigs=self.contigs, + low=80_000, + high=120_000, + base_composition=base_composition, + rng=self.rng, + ) + self.contig_sizes = { + contig: self.genome[contig].shape[0] for contig in self.contigs + } + + def init_genome_features(self): + path = self.bucket_path / self.config["GENESET_GFF3_PATH"] + path.parent.mkdir(parents=True, exist_ok=True) + simulator = Gff3Simulator( + contig_sizes=self.contig_sizes, + rng=self.rng, + gene_type="protein_coding_gene", + attrs=("Note", "description"), + ) + self.genome_features = simulator.simulate_gff(path=path) + + def write_metadata(self, release, release_path, sample_set, sequence_qc=True): + n_samples_sim = ( + self.release_manifests[release] + .set_index("sample_set") + .loc[sample_set]["sample_count"] + ) + + src_path = ( + self.fixture_dir + / "vo_aste_release_master_us_central1" + / release_path + / "metadata" + / "general" + / sample_set + / "samples.meta.csv" + ) + df_general = pd.read_csv(src_path) + df_general_ds = df_general.sample( + n_samples_sim, replace=False, random_state=self.rng + ) + samples_ds = df_general_ds["sample_id"].tolist() + dst_path = ( + self.bucket_path + / release_path + / "metadata" + / "general" + / sample_set + / "samples.meta.csv" + ) + dst_path.parent.mkdir(parents=True, exist_ok=True) + df_general_ds.to_csv(dst_path, index=False) + + if sequence_qc: + src_path = ( + self.fixture_dir + / "vo_aste_release_master_us_central1" + / release_path + / "metadata" + / "curation" + / sample_set + / "sequence_qc_stats.csv" + ) + df_sequence_qc_stats = pd.read_csv(src_path) + df_sequence_qc_stats_ds = ( + df_sequence_qc_stats.set_index("sample_id") + .loc[samples_ds] + .reset_index() + ) + dst_path = ( + self.bucket_path + / release_path + / "metadata" + / "curation" + / sample_set + / "sequence_qc_stats.csv" + ) + dst_path.parent.mkdir(parents=True, exist_ok=True) + df_sequence_qc_stats_ds.to_csv(dst_path, index=False) + + src_path = ( + self.fixture_dir + / "vo_aste_release_master_us_central1" + / release_path + / "metadata" + / "cohorts_20260402" + / sample_set + / "samples.cohorts.csv" + ) + df_coh = pd.read_csv(src_path) + df_coh_ds = df_coh.set_index("sample_id").loc[samples_ds].reset_index() + dst_path = ( + self.bucket_path + / release_path + / "metadata" + / "cohorts_20260402" + / sample_set + / "samples.cohorts.csv" + ) + dst_path.parent.mkdir(parents=True, exist_ok=True) + df_coh_ds.to_csv(dst_path, index=False) + + src_path = ( + self.fixture_dir + / "vo_aste_release_master_us_central1" + / release_path + / "metadata" + / "general" + / sample_set + / "wgs_snp_data.csv" + ) + df_cat = pd.read_csv(src_path) + df_cat_ds = df_cat.set_index("sample_id").loc[samples_ds].reset_index() + dst_path = ( + self.bucket_path + / release_path + / "metadata" + / "general" + / sample_set + / "wgs_snp_data.csv" + ) + dst_path.parent.mkdir(parents=True, exist_ok=True) + df_cat_ds.to_csv(dst_path, index=False) + + def init_metadata(self): + self.write_metadata( + release="1.0", release_path="v1.0", sample_set="1365-VO-DJ-ADBI-VMF00318" + ) + self.write_metadata( + release="1.0", release_path="v1.0", sample_set="1386-VO-KE-OCHOMO-VMF00339" + ) + self.write_metadata( + release="1.0", + release_path="v1.0", + sample_set="1367-VO-AF-DONNELLY-VMF00320", + ) + + def init_snp_sites(self): + path = self.bucket_path / "v1.0/snp_genotypes/all/sites/" + self.snp_sites, self.n_snp_sites = simulate_snp_sites( + path=path, contigs=self.contigs, genome=self.genome + ) + + def init_site_filters(self): + analysis = self.config["DEFAULT_SITE_FILTERS_ANALYSIS"] + mask = "stephensi" + p_pass = 0.59 + path = self.bucket_path / "v1.0/site_filters" / analysis / mask + simulate_site_filters( + path=path, + contigs=self.contigs, + p_pass=p_pass, + n_sites=self.n_snp_sites, + rng=self.rng, + ) + + def init_snp_genotypes(self): + for release, manifest in self.release_manifests.items(): + release_path = f"v{release}" + for rec in manifest.itertuples(): + sample_set = rec.sample_set + metadata_path = ( + self.bucket_path + / release_path + / "metadata" + / "general" + / sample_set + / "samples.meta.csv" + ) + zarr_path = ( + self.bucket_path + / release_path + / "snp_genotypes" + / "all" + / sample_set + ) + p_allele = np.array([0.981, 0.006, 0.008, 0.005]) + p_missing = np.array([0.95, 0.05]) + simulate_snp_genotypes( + zarr_path=zarr_path, + metadata_path=metadata_path, + contigs=self.contigs, + n_sites=self.n_snp_sites, + p_allele=p_allele, + p_missing=p_missing, + rng=self.rng, + ) + + def init_site_annotations(self): + path = self.bucket_path / self.config["SITE_ANNOTATIONS_ZARR_PATH"] + simulate_site_annotations(path=path, genome=self.genome, rng=self.rng) + + # For the following data fixtures we will use the "session" scope # so that the fixture data will be created only once per test # session (i.e., per invocation of pytest). @@ -3228,3 +3508,8 @@ def adir1_sim_fixture(fixture_dir): @pytest.fixture(scope="session") def amin1_sim_fixture(fixture_dir): return Amin1Simulator(fixture_dir=fixture_dir, rng=create_rng("Amin1")) + + +@pytest.fixture(scope="session") +def as1_sim_fixture(fixture_dir): + return As1Simulator(fixture_dir=fixture_dir, rng=create_rng("As1")) diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.admin_units.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.admin_units.csv new file mode 100644 index 000000000..315a9ce25 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.admin_units.csv @@ -0,0 +1,22 @@ +sample_id,country,country_ISO,country_alpha2,country_alpha3,adm1_name,adm1_ISO,admin1_geoboundaries_shape_id,admin1_representative_longitude,admin1_representative_latitude,adm2_name,admin2_geoboundaries_shape_id,admin2_representative_longitude,admin2_representative_latitude +VMF00318-0001,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0002,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0003,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0004,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0006,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0007,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0008,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0009,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0010,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0011,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0012,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0013,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0014,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0015,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0016,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0017,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0018,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0019,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0020,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0021,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 +VMF00318-0022,Djibouti,DJI,DJ,DJI,Tadjourah,DJ-TA,25444519B32938340220439,42.50876696075729,12.033220113217709,Randa,41766387B4506756798514,42.73579381029086,12.036890000170644 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.cohorts.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.cohorts.csv new file mode 100644 index 000000000..de20b8d3b --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.cohorts.csv @@ -0,0 +1,22 @@ +sample_id,country_ISO,adm1_name,adm1_ISO,adm2_name,taxon,cohort_admin1_year,cohort_admin1_month,cohort_admin1_quarter,cohort_admin2_year,cohort_admin2_month,cohort_admin2_quarter +VMF00318-0001,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0002,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0003,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0004,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0006,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0007,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0008,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0009,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0010,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0011,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0012,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0013,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0014,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0015,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0016,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0017,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0018,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0019,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0020,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0021,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 +VMF00318-0022,DJI,Tadjourah,DJ-TA,Randa,stephensi,DJ-TA_step_2023,DJ-TA_step_2023_10,DJ-TA_step_2023_Q4,DJ-TA_Randa_step_2023,DJ-TA_Randa_step_2023_10,DJ-TA_Randa_step_2023_Q4 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.taxa.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.taxa.csv new file mode 100644 index 000000000..83498da58 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1365-VO-DJ-ADBI-VMF00318/samples.taxa.csv @@ -0,0 +1,22 @@ +sample_id,taxon +VMF00318-0001,stephensi +VMF00318-0002,stephensi +VMF00318-0003,stephensi +VMF00318-0004,stephensi +VMF00318-0006,stephensi +VMF00318-0007,stephensi +VMF00318-0008,stephensi +VMF00318-0009,stephensi +VMF00318-0010,stephensi +VMF00318-0011,stephensi +VMF00318-0012,stephensi +VMF00318-0013,stephensi +VMF00318-0014,stephensi +VMF00318-0015,stephensi +VMF00318-0016,stephensi +VMF00318-0017,stephensi +VMF00318-0018,stephensi +VMF00318-0019,stephensi +VMF00318-0020,stephensi +VMF00318-0021,stephensi +VMF00318-0022,stephensi diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.admin_units.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.admin_units.csv new file mode 100644 index 000000000..f1b016e9c --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.admin_units.csv @@ -0,0 +1,25 @@ +sample_id,country,country_ISO,country_alpha2,country_alpha3,adm1_name,adm1_ISO,admin1_geoboundaries_shape_id,admin1_representative_longitude,admin1_representative_latitude,adm2_name,admin2_geoboundaries_shape_id,admin2_representative_longitude,admin2_representative_latitude +VMF00320-0001,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0003,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0004,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0005,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0006,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0007,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0008,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0009,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0010,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0012,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0013,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0014,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0015,Afghanistan,AFG,AF,AFG,Nangarhar,AF-NAN,12653393B72991944764021,70.37860309951752,34.38205294389324,Kot,17698898B22676643943482,70.58793650725207,34.150268061243054 +VMF00320-0016,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0017,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0018,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0019,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0020,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0021,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0022,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0023,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0024,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0025,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 +VMF00320-0026,Afghanistan,AFG,AF,AFG,Kunar,AF-KNR,12653393B20449218329098,71.04418840870376,35.01495569159065,Asadabad,17698898B75204567873035,71.11521147052213,34.864290650691245 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.cohorts.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.cohorts.csv new file mode 100644 index 000000000..622ca4256 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.cohorts.csv @@ -0,0 +1,25 @@ +sample_id,country_ISO,adm1_name,adm1_ISO,adm2_name,taxon,cohort_admin1_year,cohort_admin1_month,cohort_admin1_quarter,cohort_admin2_year,cohort_admin2_month,cohort_admin2_quarter +VMF00320-0001,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0003,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0004,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0005,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0006,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0007,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0008,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0009,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0010,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0012,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0013,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0014,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0015,AFG,Nangarhar,AF-NAN,Kot,stephensi,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017,AF-NAN_Kot_step_2017 +VMF00320-0016,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0017,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0018,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0019,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0020,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0021,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0022,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0023,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0024,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0025,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 +VMF00320-0026,AFG,Kunar,AF-KNR,Asadabad,stephensi,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017,AF-KNR_Asadabad_step_2017 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.taxa.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.taxa.csv new file mode 100644 index 000000000..b8f71f6d0 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1367-VO-AF-DONNELLY-VMF00320/samples.taxa.csv @@ -0,0 +1,25 @@ +sample_id,taxon +VMF00320-0001,stephensi +VMF00320-0003,stephensi +VMF00320-0004,stephensi +VMF00320-0005,stephensi +VMF00320-0006,stephensi +VMF00320-0007,stephensi +VMF00320-0008,stephensi +VMF00320-0009,stephensi +VMF00320-0010,stephensi +VMF00320-0012,stephensi +VMF00320-0013,stephensi +VMF00320-0014,stephensi +VMF00320-0015,stephensi +VMF00320-0016,stephensi +VMF00320-0017,stephensi +VMF00320-0018,stephensi +VMF00320-0019,stephensi +VMF00320-0020,stephensi +VMF00320-0021,stephensi +VMF00320-0022,stephensi +VMF00320-0023,stephensi +VMF00320-0024,stephensi +VMF00320-0025,stephensi +VMF00320-0026,stephensi diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.admin_units.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.admin_units.csv new file mode 100644 index 000000000..4543052d6 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.admin_units.csv @@ -0,0 +1,30 @@ +sample_id,country,country_ISO,country_alpha2,country_alpha3,adm1_name,adm1_ISO,admin1_geoboundaries_shape_id,admin1_representative_longitude,admin1_representative_latitude,adm2_name,admin2_geoboundaries_shape_id,admin2_representative_longitude,admin2_representative_latitude +VMF00339-0001,Kenya,KEN,KE,KEN,Marsabit,KE-25,32016919B18178000212460,37.71235874316875,2.86021590254191,Saku,3690345B62375785274642,37.949602359866944,2.376314282500061 +VMF00339-0002,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0003,Kenya,KEN,KE,KEN,Isiolo,KE-09,32016919B25263975567841,38.73790706541679,1.042712450619149,Isiolo North,3690345B97238086520982,38.72868612295035,1.4409881830000302 +VMF00339-0004,Kenya,KEN,KE,KEN,Isiolo,KE-09,32016919B25263975567841,38.73790706541679,1.042712450619149,Isiolo North,3690345B97238086520982,38.72868612295035,1.4409881830000302 +VMF00339-0005,Kenya,KEN,KE,KEN,Isiolo,KE-09,32016919B25263975567841,38.73790706541679,1.042712450619149,Isiolo North,3690345B97238086520982,38.72868612295035,1.4409881830000302 +VMF00339-0006,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0007,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0008,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0009,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0010,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0011,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0012,Kenya,KEN,KE,KEN,Isiolo,KE-09,32016919B25263975567841,38.73790706541679,1.042712450619149,Isiolo North,3690345B97238086520982,38.72868612295035,1.4409881830000302 +VMF00339-0013,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0014,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0015,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0016,Kenya,KEN,KE,KEN,Isiolo,KE-09,32016919B25263975567841,38.73790706541679,1.042712450619149,Isiolo North,3690345B97238086520982,38.72868612295035,1.4409881830000302 +VMF00339-0017,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0018,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0019,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0020,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0021,Kenya,KEN,KE,KEN,Isiolo,KE-09,32016919B25263975567841,38.73790706541679,1.042712450619149,Isiolo North,3690345B97238086520982,38.72868612295035,1.4409881830000302 +VMF00339-0022,Kenya,KEN,KE,KEN,Mandera,KE-24,32016919B58193784942652,40.664688622614925,3.281521320352681,Mandera East,3690345B96575375377195,41.501296194606724,3.6839683165000565 +VMF00339-0025,Kenya,KEN,KE,KEN,Samburu,KE-37,32016919B45231166049088,36.93963953208977,1.541152596665057,Samburu East,3690345B42548654365756,37.53498951079123,1.109154637500069 +VMF00339-0026,Kenya,KEN,KE,KEN,Samburu,KE-37,32016919B45231166049088,36.93963953208977,1.541152596665057,Samburu East,3690345B42548654365756,37.53498951079123,1.109154637500069 +VMF00339-0027,Kenya,KEN,KE,KEN,Samburu,KE-37,32016919B45231166049088,36.93963953208977,1.541152596665057,Samburu East,3690345B42548654365756,37.53498951079123,1.109154637500069 +VMF00339-0028,Kenya,KEN,KE,KEN,Samburu,KE-37,32016919B45231166049088,36.93963953208977,1.541152596665057,Samburu East,3690345B42548654365756,37.53498951079123,1.109154637500069 +VMF00339-0029,Kenya,KEN,KE,KEN,Samburu,KE-37,32016919B45231166049088,36.93963953208977,1.541152596665057,Samburu East,3690345B42548654365756,37.53498951079123,1.109154637500069 +VMF00339-0030,Kenya,KEN,KE,KEN,Samburu,KE-37,32016919B45231166049088,36.93963953208977,1.541152596665057,Samburu East,3690345B42548654365756,37.53498951079123,1.109154637500069 +VMF00339-0031,Kenya,KEN,KE,KEN,Samburu,KE-37,32016919B45231166049088,36.93963953208977,1.541152596665057,Samburu East,3690345B42548654365756,37.53498951079123,1.109154637500069 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.cohorts.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.cohorts.csv new file mode 100644 index 000000000..832e52f43 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.cohorts.csv @@ -0,0 +1,30 @@ +sample_id,country_ISO,adm1_name,adm1_ISO,adm2_name,taxon,cohort_admin1_year,cohort_admin1_month,cohort_admin1_quarter,cohort_admin2_year,cohort_admin2_month,cohort_admin2_quarter +VMF00339-0001,KEN,Marsabit,KE-25,Saku,stephensi,KE-25_step_2022,KE-25_step_2022_12,KE-25_step_2022_Q4,KE-25_Saku_step_2022,KE-25_Saku_step_2022_12,KE-25_Saku_step_2022_Q4 +VMF00339-0002,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0003,KEN,Isiolo,KE-09,Isiolo North,stephensi,KE-09_step_2024,KE-09_step_2024,KE-09_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024 +VMF00339-0004,KEN,Isiolo,KE-09,Isiolo North,stephensi,KE-09_step_2024,KE-09_step_2024,KE-09_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024 +VMF00339-0005,KEN,Isiolo,KE-09,Isiolo North,stephensi,KE-09_step_2024,KE-09_step_2024,KE-09_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024 +VMF00339-0006,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0007,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0008,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0009,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0010,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0011,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0012,KEN,Isiolo,KE-09,Isiolo North,stephensi,KE-09_step_2024,KE-09_step_2024,KE-09_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024 +VMF00339-0013,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0014,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0015,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0016,KEN,Isiolo,KE-09,Isiolo North,stephensi,KE-09_step_2024,KE-09_step_2024,KE-09_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024 +VMF00339-0017,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0018,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0019,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0020,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0021,KEN,Isiolo,KE-09,Isiolo North,stephensi,KE-09_step_2024,KE-09_step_2024,KE-09_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024,KE-09_Isiolo-North_step_2024 +VMF00339-0022,KEN,Mandera,KE-24,Mandera East,stephensi,KE-24_step_2024,KE-24_step_2024,KE-24_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024,KE-24_Mandera-East_step_2024 +VMF00339-0025,KEN,Samburu,KE-37,Samburu East,stephensi,KE-37_step_2024,KE-37_step_2024,KE-37_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024 +VMF00339-0026,KEN,Samburu,KE-37,Samburu East,stephensi,KE-37_step_2024,KE-37_step_2024,KE-37_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024 +VMF00339-0027,KEN,Samburu,KE-37,Samburu East,stephensi,KE-37_step_2024,KE-37_step_2024,KE-37_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024 +VMF00339-0028,KEN,Samburu,KE-37,Samburu East,stephensi,KE-37_step_2024,KE-37_step_2024,KE-37_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024 +VMF00339-0029,KEN,Samburu,KE-37,Samburu East,stephensi,KE-37_step_2024,KE-37_step_2024,KE-37_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024 +VMF00339-0030,KEN,Samburu,KE-37,Samburu East,stephensi,KE-37_step_2024,KE-37_step_2024,KE-37_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024 +VMF00339-0031,KEN,Samburu,KE-37,Samburu East,stephensi,KE-37_step_2024,KE-37_step_2024,KE-37_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024,KE-37_Samburu-East_step_2024 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.taxa.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.taxa.csv new file mode 100644 index 000000000..ca1a07dd6 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/cohorts_20260402/1386-VO-KE-OCHOMO-VMF00339/samples.taxa.csv @@ -0,0 +1,30 @@ +sample_id,taxon +VMF00339-0001,stephensi +VMF00339-0002,stephensi +VMF00339-0003,stephensi +VMF00339-0004,stephensi +VMF00339-0005,stephensi +VMF00339-0006,stephensi +VMF00339-0007,stephensi +VMF00339-0008,stephensi +VMF00339-0009,stephensi +VMF00339-0010,stephensi +VMF00339-0011,stephensi +VMF00339-0012,stephensi +VMF00339-0013,stephensi +VMF00339-0014,stephensi +VMF00339-0015,stephensi +VMF00339-0016,stephensi +VMF00339-0017,stephensi +VMF00339-0018,stephensi +VMF00339-0019,stephensi +VMF00339-0020,stephensi +VMF00339-0021,stephensi +VMF00339-0022,stephensi +VMF00339-0025,stephensi +VMF00339-0026,stephensi +VMF00339-0027,stephensi +VMF00339-0028,stephensi +VMF00339-0029,stephensi +VMF00339-0030,stephensi +VMF00339-0031,stephensi diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1365-VO-DJ-ADBI-VMF00318/sequence_qc_stats.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1365-VO-DJ-ADBI-VMF00318/sequence_qc_stats.csv new file mode 100644 index 000000000..9c225197a --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1365-VO-DJ-ADBI-VMF00318/sequence_qc_stats.csv @@ -0,0 +1,22 @@ +sample_id,mean_cov,median_cov,modal_cov,mean_cov_2RL,median_cov_2RL,mode_cov_2RL,mean_cov_3RL,median_cov_3RL,mode_cov_3RL,mean_cov_X,median_cov_X,mode_cov_X,frac_gen_cov,divergence,contam_pct,contam_LLR +VMF00318-0001,21.39,21,20,21.44,21,20,21.11,21,20,22.34,21,20,0.953,0.01965,, +VMF00318-0002,26.37,26,25,26.45,26,25,26.16,26,25,26.88,25,25,0.954,0.01949,, +VMF00318-0003,40.61,40,40,40.83,40,40,40.25,40,40,41.17,39,39,0.954,0.01927,, +VMF00318-0004,22.26,22,21,22.34,22,21,22.09,22,21,22.61,21,21,0.953,0.0195,, +VMF00318-0006,32.69,32,32,32.78,32,32,32.39,32,32,33.58,32,31,0.953,0.01936,, +VMF00318-0007,27.86,27,27,27.94,27,27,27.47,27,27,29.1,27,26,0.955,0.01939,, +VMF00318-0008,25.07,24,24,25.18,25,24,24.82,24,24,25.57,24,24,0.953,0.01942,, +VMF00318-0009,30.95,30,30,31.03,30,30,30.67,30,30,31.73,30,29,0.953,0.01939,, +VMF00318-0010,26.33,26,25,26.46,26,25,26.07,26,25,26.85,26,25,0.952,0.01945,, +VMF00318-0011,27.77,27,27,27.89,27,27,27.59,27,27,28.0,27,26,0.953,0.01946,, +VMF00318-0012,38.43,38,38,38.6,38,38,38.16,38,38,38.73,37,37,0.955,0.01929,, +VMF00318-0013,26.28,26,25,26.4,26,25,25.96,25,25,27.1,25,25,0.953,0.01949,, +VMF00318-0014,22.27,22,21,22.41,22,21,22.04,22,21,22.6,21,21,0.953,0.01962,, +VMF00318-0015,26.28,26,25,26.44,26,25,26.0,26,25,26.77,25,25,0.954,0.01935,, +VMF00318-0016,24.79,24,24,24.87,24,24,24.51,24,24,25.56,24,24,0.953,0.01952,, +VMF00318-0017,19.94,19,19,20.17,19,19,19.67,19,19,20.08,19,19,0.952,0.01988,, +VMF00318-0018,27.63,27,26,27.77,27,27,27.34,27,26,28.25,27,26,0.954,0.01939,, +VMF00318-0019,25.8,25,25,26.04,25,25,25.53,25,25,25.93,25,24,0.952,0.01945,, +VMF00318-0020,25.39,25,24,25.52,25,24,25.14,25,24,25.91,24,24,0.952,0.01952,, +VMF00318-0021,23.72,23,23,23.81,23,23,23.47,23,23,24.37,23,23,0.953,0.0196,, +VMF00318-0022,22.61,22,21,22.81,22,22,22.37,22,21,22.76,22,21,0.952,0.0196,, diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1367-VO-AF-DONNELLY-VMF00320/sequence_qc_stats.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1367-VO-AF-DONNELLY-VMF00320/sequence_qc_stats.csv new file mode 100644 index 000000000..f80e4dce7 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1367-VO-AF-DONNELLY-VMF00320/sequence_qc_stats.csv @@ -0,0 +1,25 @@ +sample_id,mean_cov,median_cov,modal_cov,mean_cov_2RL,median_cov_2RL,mode_cov_2RL,mean_cov_3RL,median_cov_3RL,mode_cov_3RL,mean_cov_X,median_cov_X,mode_cov_X,frac_gen_cov,divergence,contam_pct,contam_LLR +VMF00320-0001,25.23,25,24,25.29,25,24,24.92,24,24,26.27,25,24,0.951,0.0201,, +VMF00320-0003,19.5,19,18,19.56,19,18,19.22,19,18,20.36,19,18,0.95,0.02057,, +VMF00320-0004,26.65,26,26,26.74,26,26,26.37,26,26,27.38,26,25,0.952,0.02001,, +VMF00320-0005,22.31,21,21,22.45,22,21,21.92,21,21,23.28,21,21,0.952,0.02019,, +VMF00320-0006,21.82,21,21,21.9,21,21,21.46,21,21,22.99,21,21,0.951,0.0202,, +VMF00320-0007,22.83,22,22,22.93,22,22,22.51,22,22,23.71,22,22,0.953,0.02,, +VMF00320-0008,20.07,19,19,20.14,19,19,19.79,19,19,20.91,19,19,0.953,0.02024,, +VMF00320-0009,22.45,22,22,22.53,22,22,22.13,22,21,23.36,22,21,0.953,0.02008,, +VMF00320-0010,24.74,24,24,24.89,24,24,24.37,24,24,25.61,24,24,0.952,0.02005,, +VMF00320-0012,23.1,22,22,23.15,22,22,22.74,22,22,24.31,22,22,0.954,0.02002,, +VMF00320-0013,24.23,23,23,24.23,24,23,23.82,23,23,25.92,24,23,0.952,0.02003,, +VMF00320-0014,24.33,24,23,24.46,24,23,23.94,24,23,25.31,24,23,0.953,0.01994,, +VMF00320-0015,23.78,23,23,23.84,23,23,23.43,23,23,24.94,23,23,0.951,0.02015,, +VMF00320-0016,22.06,21,21,22.11,21,21,21.8,21,21,22.91,21,21,0.953,0.02009,, +VMF00320-0017,28.2,28,27,28.32,28,27,27.87,28,27,29.09,27,27,0.953,0.01995,, +VMF00320-0018,43.42,43,43,43.51,43,43,43.02,43,43,44.65,43,42,0.955,0.01984,, +VMF00320-0019,19.93,19,18,20.07,19,19,19.64,19,18,20.52,19,18,0.952,0.02033,, +VMF00320-0020,22.89,22,22,22.96,22,22,22.61,22,22,23.7,22,22,0.951,0.02024,, +VMF00320-0021,23.32,22,22,23.53,23,22,22.96,22,22,23.89,22,22,0.951,0.02028,, +VMF00320-0022,27.43,27,27,27.53,27,27,27.09,27,27,28.4,27,26,0.952,0.01998,, +VMF00320-0023,24.37,24,23,24.39,24,23,24.02,24,23,25.7,24,23,0.954,0.01997,, +VMF00320-0024,22.3,22,21,22.41,22,21,22.08,21,21,22.76,21,21,0.951,0.02017,, +VMF00320-0025,19.25,18,18,19.38,18,18,18.86,18,18,20.27,19,18,0.952,0.02052,, +VMF00320-0026,20.41,20,19,20.45,20,19,20.08,20,19,21.54,20,19,0.952,0.02036,, diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1386-VO-KE-OCHOMO-VMF00339/sequence_qc_stats.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1386-VO-KE-OCHOMO-VMF00339/sequence_qc_stats.csv new file mode 100644 index 000000000..180cde55c --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/curation/1386-VO-KE-OCHOMO-VMF00339/sequence_qc_stats.csv @@ -0,0 +1,30 @@ +sample_id,mean_cov,median_cov,modal_cov,mean_cov_2RL,median_cov_2RL,mode_cov_2RL,mean_cov_3RL,median_cov_3RL,mode_cov_3RL,mean_cov_X,median_cov_X,mode_cov_X,frac_gen_cov,divergence,contam_pct,contam_LLR +VMF00339-0001,17.37,16,14,17.38,16,14,17.11,16,14,18.42,15,13,0.946,0.02138,, +VMF00339-0002,24.61,24,24,24.74,24,24,24.37,24,24,24.97,24,23,0.953,0.01913,, +VMF00339-0003,28.79,28,28,29.0,28,28,28.58,28,28,28.77,28,28,0.953,0.01901,, +VMF00339-0004,28.2,28,27,28.38,28,28,28.04,28,27,28.08,27,27,0.952,0.01922,, +VMF00339-0005,22.43,22,22,23.57,23,23,23.16,23,22,14.47,12,11,0.95,0.01924,, +VMF00339-0006,17.63,17,17,18.55,18,17,18.2,18,17,11.34,9,8,0.951,0.01932,, +VMF00339-0007,22.79,22,23,23.93,23,23,23.5,23,23,14.87,12,11,0.951,0.01928,, +VMF00339-0008,21.8,21,22,22.94,22,22,22.43,22,22,14.26,11,10,0.952,0.01928,, +VMF00339-0009,19.93,19,20,20.95,20,20,20.53,20,20,13.0,10,9,0.95,0.01927,, +VMF00339-0010,22.39,22,22,23.55,23,22,23.04,23,22,14.63,12,11,0.951,0.01925,, +VMF00339-0011,23.43,23,23,24.62,24,23,24.17,24,23,15.21,12,11,0.95,0.0193,, +VMF00339-0012,25.73,25,26,27.09,26,26,26.67,26,26,16.0,13,12,0.951,0.01917,, +VMF00339-0013,23.46,23,24,24.63,24,24,24.33,24,24,14.79,12,12,0.953,0.01913,, +VMF00339-0014,22.16,22,22,23.32,23,22,22.87,22,22,14.22,12,11,0.951,0.01925,, +VMF00339-0015,23.42,23,23,24.64,24,24,24.16,24,23,15.1,12,11,0.951,0.01925,, +VMF00339-0016,22.75,22,22,22.88,22,22,22.53,22,22,23.08,21,20,0.955,0.01911,, +VMF00339-0017,26.5,26,27,27.89,27,27,27.37,27,27,16.9,14,13,0.951,0.01925,, +VMF00339-0018,22.36,22,22,22.53,22,22,22.13,22,22,22.61,21,21,0.952,0.01908,, +VMF00339-0019,28.62,28,28,28.78,28,28,28.33,28,28,29.12,28,28,0.953,0.01914,, +VMF00339-0020,22.78,22,23,24.03,23,23,23.5,23,23,14.35,12,11,0.951,0.01919,, +VMF00339-0021,27.09,27,26,27.23,27,27,26.83,27,26,27.54,26,26,0.953,0.01919,, +VMF00339-0022,24.24,24,24,24.39,24,24,23.94,24,24,24.86,24,24,0.952,0.01911,, +VMF00339-0025,21.49,21,21,22.65,22,22,22.15,22,21,13.77,11,10,0.95,0.0193,, +VMF00339-0026,24.5,24,25,25.8,25,25,25.22,25,25,15.97,13,12,0.951,0.01923,, +VMF00339-0027,22.08,22,22,23.3,23,22,22.74,22,22,14.06,12,11,0.951,0.01931,, +VMF00339-0028,29.04,29,28,29.18,29,28,28.81,29,28,29.42,28,28,0.952,0.01918,, +VMF00339-0029,23.81,23,24,25.03,24,24,24.6,24,24,15.31,12,11,0.951,0.01933,, +VMF00339-0030,23.05,23,23,24.24,23,23,23.78,23,23,14.87,12,11,0.951,0.01928,, +VMF00339-0031,23.85,23,24,25.12,24,24,24.61,24,24,15.21,12,12,0.951,0.01928,, diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/samples.meta.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/samples.meta.csv new file mode 100644 index 000000000..2e44fc1d5 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/samples.meta.csv @@ -0,0 +1,22 @@ +sample_id,partner_sample_id,contributor,country,location,year,month,latitude,longitude,sex_call +VMF00318-0001,Djib_F_01,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0002,Djib_F_03,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0003,Djib_F_04,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0004,Djib_F_05,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0006,Djib_F_07,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0007,Djib_F_08,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0008,Djib_F_10,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0009,Djib_F_11,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0010,Djib_F_12,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0011,Djib_F_13,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0012,Djib_F_15,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0013,Djib_F_16,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0014,Djib_F_17,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0015,Djib_F_19,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0016,Djib_F_21,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0017,Djib_F_22,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0018,Djib_F_24,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0019,Djib_F_26,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0020,Djib_F_27,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0021,Djib_F_28,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F +VMF00318-0022,Djib_F_29,Bouh Adbi,Djibouti,DjiboutiCity,2023,10,11.825,42.590,F diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/sequence_qc_stats.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/sequence_qc_stats.csv new file mode 100644 index 000000000..f2bf9fb68 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/sequence_qc_stats.csv @@ -0,0 +1,22 @@ +sample_id,mean_cov,median_cov,modal_cov,mean_cov_2RL,median_cov_2RL,mode_cov_2RL,mean_cov_3RL,median_cov_3RL,mode_cov_3RL,mean_cov_X,median_cov_X,mode_cov_X,frac_gen_cov,divergence,contam_pct,contam_LLR +VMF00318-0001,21.39,21,20,21.44,21,20,21.11,21,20,22.34,21,20,0.953,0.01965,0.0,0.0 +VMF00318-0002,26.37,26,25,26.45,26,25,26.16,26,25,26.88,25,25,0.954,0.01949,0.0,0.0 +VMF00318-0003,40.61,40,40,40.83,40,40,40.25,40,40,41.17,39,39,0.954,0.01927,0.0,0.0 +VMF00318-0004,22.26,22,21,22.34,22,21,22.09,22,21,22.61,21,21,0.953,0.0195,0.0,0.0 +VMF00318-0006,32.69,32,32,32.78,32,32,32.39,32,32,33.58,32,31,0.953,0.01936,0.0,0.0 +VMF00318-0007,27.86,27,27,27.94,27,27,27.47,27,27,29.1,27,26,0.955,0.01939,0.0,0.0 +VMF00318-0008,25.07,24,24,25.18,25,24,24.82,24,24,25.57,24,24,0.953,0.01942,0.0,0.0 +VMF00318-0009,30.95,30,30,31.03,30,30,30.67,30,30,31.73,30,29,0.953,0.01939,0.0,0.0 +VMF00318-0010,26.33,26,25,26.46,26,25,26.07,26,25,26.85,26,25,0.952,0.01945,0.0,0.0 +VMF00318-0011,27.77,27,27,27.89,27,27,27.59,27,27,28.0,27,26,0.953,0.01946,0.0,0.0 +VMF00318-0012,38.43,38,38,38.6,38,38,38.16,38,38,38.73,37,37,0.955,0.01929,0.0,0.0 +VMF00318-0013,26.28,26,25,26.4,26,25,25.96,25,25,27.1,25,25,0.953,0.01949,0.0,0.0 +VMF00318-0014,22.27,22,21,22.41,22,21,22.04,22,21,22.6,21,21,0.953,0.01962,0.0,0.0 +VMF00318-0015,26.28,26,25,26.44,26,25,26.0,26,25,26.77,25,25,0.954,0.01935,0.0,0.0 +VMF00318-0016,24.79,24,24,24.87,24,24,24.51,24,24,25.56,24,24,0.953,0.01952,0.0,0.0 +VMF00318-0017,19.94,19,19,20.17,19,19,19.67,19,19,20.08,19,19,0.952,0.01988,0.0,0.0 +VMF00318-0018,27.63,27,26,27.77,27,27,27.34,27,26,28.25,27,26,0.954,0.01939,0.0,0.0 +VMF00318-0019,25.8,25,25,26.04,25,25,25.53,25,25,25.93,25,24,0.952,0.01945,0.0,0.0 +VMF00318-0020,25.39,25,24,25.52,25,24,25.14,25,24,25.91,24,24,0.952,0.01952,0.0,0.0 +VMF00318-0021,23.72,23,23,23.81,23,23,23.47,23,23,24.37,23,23,0.953,0.0196,0.0,0.0 +VMF00318-0022,22.61,22,21,22.81,22,22,22.37,22,21,22.76,22,21,0.952,0.0196,0.0,0.0 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/wgs_snp_data.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/wgs_snp_data.csv new file mode 100644 index 000000000..a5f146d42 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1365-VO-DJ-ADBI-VMF00318/wgs_snp_data.csv @@ -0,0 +1,22 @@ +sample_id,alignments_bam,alignments_bam_md5,snp_genotypes_vcf,snp_genotypes_vcf_md5,snp_genotypes_zarr,snp_genotypes_zarr_md5,pipeline_version +VMF00318-0001,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0001.fixmate.bam,a4b4294fcbfa47f14748473484bee7fd,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0001.vcf.gz,1dae72262d4b80ea6a811ddaa8465c64,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0001.zarr.zip,dcdc953f0e2978ca482198bc9fa9a25a,v1.1 +VMF00318-0002,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0002.fixmate.bam,647ee77935d58d288d85685be26ebce1,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0002.vcf.gz,28546db8eb49f80b6796a4158c50da47,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0002.zarr.zip,34d77c540f82769fde86526a6d4d5374,v1.1 +VMF00318-0003,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0003.fixmate.bam,629c237991cd8fa4db1483ef6ad1b4ad,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0003.vcf.gz,d979df5d8a59be57448d7cfe5622c164,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0003.zarr.zip,c947cde32f89091a9cde3b77ca968b99,v1.1 +VMF00318-0004,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0004.fixmate.bam,315520aada2accd611e4c4d890c40aaf,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0004.vcf.gz,648a2818f5b90b00cbd0087ec6019a1c,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0004.zarr.zip,b2e53cc6e736a57ca3f6cf9389a6b910,v1.1 +VMF00318-0006,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0006.fixmate.bam,afc1dea2fc8c69ca10890ad15339abd5,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0006.vcf.gz,e394a787be59bd2ffd52add9217ae241,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0006.zarr.zip,581ebdfaab4ece80c2fb6dac3fefd108,v1.1 +VMF00318-0007,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0007.fixmate.bam,fd2cafed6fac3eb8c7128e2403cf5398,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0007.vcf.gz,87ebf5d57f50e483353f1a68eaaa3919,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0007.zarr.zip,90540c81c20f66f70ddc767679d42a29,v1.1 +VMF00318-0008,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0008.fixmate.bam,4c0afcd9d2b08f17a2bc8606ca90a1b7,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0008.vcf.gz,e3dbab84c1a1761150445861e79a401d,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0008.zarr.zip,3a27bd6d2cfd6413b1ecea7c9489b5b8,v1.1 +VMF00318-0009,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0009.fixmate.bam,c0a53b30288d0f4f6a34889c032192fd,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0009.vcf.gz,3cc4da186350206bf9800c1a3bc67112,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0009.zarr.zip,e6d737add473ef9041bb2939214362e8,v1.1 +VMF00318-0010,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0010.fixmate.bam,6b13742490c0e4e79c29bd48a0e16de0,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0010.vcf.gz,8e15df9f8c872bd99250212b823ba541,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0010.zarr.zip,619fa1f7f362435c282581734ade44ca,v1.1 +VMF00318-0011,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0011.fixmate.bam,fdf6d22394757b0769fc1ffccde05983,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0011.vcf.gz,14c37935e0b9fdfaef37b208dc24cf56,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0011.zarr.zip,480f9742986d0b36160aa78ffafaa290,v1.1 +VMF00318-0012,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0012.fixmate.bam,04a6657841a6fb19ceeb925c0d4d14d3,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0012.vcf.gz,2a1a3a7eec9d989ac8a7ae104dbfb1e3,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0012.zarr.zip,31f7a4d11604974fa003659898e5301c,v1.1 +VMF00318-0013,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0013.fixmate.bam,9471a385670684734e5dc45598c7dc9a,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0013.vcf.gz,89625a9415e9c491c176fc41c540d9b2,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0013.zarr.zip,2d06d69ae57a326771b291eab3ead9e0,v1.1 +VMF00318-0014,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0014.fixmate.bam,6ac54d01695c5beeb60d3b30b8224c6a,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0014.vcf.gz,fc573d5c24efe036403b9cb8b58d45ad,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0014.zarr.zip,1c35b756b06fd7efd605ef59a6249caf,v1.1 +VMF00318-0015,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0015.fixmate.bam,096358886c1c8240afd6e75aece77bd6,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0015.vcf.gz,d7604fd2f6eed3ea8acce1e6ea344ad6,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0015.zarr.zip,3340ff7714566b87ae2c3e73a64486fa,v1.1 +VMF00318-0016,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0016.fixmate.bam,c7edf1acc54524c83e3a3b2b17695fb6,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0016.vcf.gz,fefbad959597498b56234b119edd2c64,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0016.zarr.zip,e8e6d0f4a9b6839c2fb214edfa99e134,v1.1 +VMF00318-0017,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0017.fixmate.bam,a10f97e8f49c23e28eed055943cc229e,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0017.vcf.gz,0e2027dded47262bea558516a11d394f,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0017.zarr.zip,b6d50c16fd544319c3245e9f997a2277,v1.1 +VMF00318-0018,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0018.fixmate.bam,afb5a130e659b4f4b5c6e556a2fbf800,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0018.vcf.gz,88b42278fc2d8550bf389881218c9943,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0018.zarr.zip,465a0fb705cd42738fb48a2bf3daa277,v1.1 +VMF00318-0019,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0019.fixmate.bam,e322a536af2f8f430e56d96ef52a1654,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0019.vcf.gz,2c9a332c25e2fee0f743017d434a8f4e,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0019.zarr.zip,c5d7f7fac18766aba4c4ce6685a8af33,v1.1 +VMF00318-0020,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0020.fixmate.bam,4d8ba28309cd17a3edd101fd362491ec,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0020.vcf.gz,c5164c67610c6a4591da5f29305a12ae,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0020.zarr.zip,c9817ad959f34bba8bb6899fc819c019,v1.1 +VMF00318-0021,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0021.fixmate.bam,caa2bfd22e105e546d04e9da29b3f619,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0021.vcf.gz,65cf15a869925f3b927d097d0a13c541,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0021.zarr.zip,805a137ec8e52d9a522a73395ff0bf32,v1.1 +VMF00318-0022,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0022.fixmate.bam,9964c9a7fe3336ce9679fbf7038bd153,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0022.vcf.gz,155405f1043c3d2081ddfece96bf5670,https://1365-vo-dj-adbi-vmf00318-aste1.cog.sanger.ac.uk/VMF00318-0022.zarr.zip,953972fb259768b5dbea06b3fd3b9fbd,v1.1 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/samples.meta.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/samples.meta.csv new file mode 100644 index 000000000..a7d296a0e --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/samples.meta.csv @@ -0,0 +1,25 @@ +sample_id,partner_sample_id,contributor,country,location,year,month,latitude,longitude,sex_call +VMF00320-0001,AfNA1,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0003,AfNC4,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0004,AfNC7,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0005,AfND7,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0006,AfND8,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0007,AfNE1,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0008,AfNE6,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0009,AfNE7,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0010,AfNF1,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0012,AfNG1,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0013,AfNG8,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0014,AfNH7,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0015,AfNH8,Martin Donnelly,Afghanistan,Nangarhar,2017,-1,34.172,70.622,F +VMF00320-0016,AfkB9,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0017,AfkC4,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0018,AfkC5,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0019,AfkE3,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0020,AfkE4,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0021,AfkE5,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0022,AfkF3,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0023,AfkF4,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0024,AfkF7,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0025,AfkH3,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F +VMF00320-0026,AfkH4,Martin Donnelly,Afghanistan,Kunar,2017,-1,34.847,71.097,F diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/sequence_qc_stats.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/sequence_qc_stats.csv new file mode 100644 index 000000000..ec4f5e68b --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/sequence_qc_stats.csv @@ -0,0 +1,25 @@ +sample_id,mean_cov,median_cov,modal_cov,mean_cov_2RL,median_cov_2RL,mode_cov_2RL,mean_cov_3RL,median_cov_3RL,mode_cov_3RL,mean_cov_X,median_cov_X,mode_cov_X,frac_gen_cov,divergence,contam_pct,contam_LLR +VMF00320-0001,25.23,25,24,25.29,25,24,24.92,24,24,26.27,25,24,0.951,0.0201,0.0,0.0 +VMF00320-0003,19.5,19,18,19.56,19,18,19.22,19,18,20.36,19,18,0.95,0.02057,0.0,0.0 +VMF00320-0004,26.65,26,26,26.74,26,26,26.37,26,26,27.38,26,25,0.952,0.02001,0.0,0.0 +VMF00320-0005,22.31,21,21,22.45,22,21,21.92,21,21,23.28,21,21,0.952,0.02019,0.0,0.0 +VMF00320-0006,21.82,21,21,21.9,21,21,21.46,21,21,22.99,21,21,0.951,0.0202,0.0,0.0 +VMF00320-0007,22.83,22,22,22.93,22,22,22.51,22,22,23.71,22,22,0.953,0.02,0.0,0.0 +VMF00320-0008,20.07,19,19,20.14,19,19,19.79,19,19,20.91,19,19,0.953,0.02024,0.0,0.0 +VMF00320-0009,22.45,22,22,22.53,22,22,22.13,22,21,23.36,22,21,0.953,0.02008,0.0,0.0 +VMF00320-0010,24.74,24,24,24.89,24,24,24.37,24,24,25.61,24,24,0.952,0.02005,0.0,0.0 +VMF00320-0012,23.1,22,22,23.15,22,22,22.74,22,22,24.31,22,22,0.954,0.02002,0.0,0.0 +VMF00320-0013,24.23,23,23,24.23,24,23,23.82,23,23,25.92,24,23,0.952,0.02003,0.0,0.0 +VMF00320-0014,24.33,24,23,24.46,24,23,23.94,24,23,25.31,24,23,0.953,0.01994,0.0,0.0 +VMF00320-0015,23.78,23,23,23.84,23,23,23.43,23,23,24.94,23,23,0.951,0.02015,0.0,0.0 +VMF00320-0016,22.06,21,21,22.11,21,21,21.8,21,21,22.91,21,21,0.953,0.02009,0.0,0.0 +VMF00320-0017,28.2,28,27,28.32,28,27,27.87,28,27,29.09,27,27,0.953,0.01995,0.0,0.0 +VMF00320-0018,43.42,43,43,43.51,43,43,43.02,43,43,44.65,43,42,0.955,0.01984,0.0,0.0 +VMF00320-0019,19.93,19,18,20.07,19,19,19.64,19,18,20.52,19,18,0.952,0.02033,0.0,0.0 +VMF00320-0020,22.89,22,22,22.96,22,22,22.61,22,22,23.7,22,22,0.951,0.02024,0.0,0.0 +VMF00320-0021,23.32,22,22,23.53,23,22,22.96,22,22,23.89,22,22,0.951,0.02028,0.0,0.0 +VMF00320-0022,27.43,27,27,27.53,27,27,27.09,27,27,28.4,27,26,0.952,0.01998,0.0,0.0 +VMF00320-0023,24.37,24,23,24.39,24,23,24.02,24,23,25.7,24,23,0.954,0.01997,0.0,0.0 +VMF00320-0024,22.3,22,21,22.41,22,21,22.08,21,21,22.76,21,21,0.951,0.02017,0.0,0.0 +VMF00320-0025,19.25,18,18,19.38,18,18,18.86,18,18,20.27,19,18,0.952,0.02052,0.0,0.0 +VMF00320-0026,20.41,20,19,20.45,20,19,20.08,20,19,21.54,20,19,0.952,0.02036,0.0,0.0 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/wgs_snp_data.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/wgs_snp_data.csv new file mode 100644 index 000000000..905d0ea55 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1367-VO-AF-DONNELLY-VMF00320/wgs_snp_data.csv @@ -0,0 +1,25 @@ +sample_id,alignments_bam,alignments_bam_md5,snp_genotypes_vcf,snp_genotypes_vcf_md5,snp_genotypes_zarr,snp_genotypes_zarr_md5,pipeline_version +VMF00320-0001,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0001.fixmate.bam,7d90cbaed6cc5f6760f17e1ea5b30b61,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0001.vcf.gz,10b10566228e5fea3cb707e712caa954,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0001.zarr.zip,c6904e33b7608e9d48c608f4290c8e83,v1.1 +VMF00320-0003,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0003.fixmate.bam,0424d77ad9c83c0509b227fa00506be9,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0003.vcf.gz,87a9b0520d54c3d0a049a82cba99f64e,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0003.zarr.zip,b50ce3209e56fd2aa064eaeb38db4ad0,v1.1 +VMF00320-0004,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0004.fixmate.bam,e9720520e53101c4062de5fe0b62bb77,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0004.vcf.gz,7a5d5c4b7148dea927e80080ebc4b7d9,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0004.zarr.zip,8c91fe633ddaa23391a264f22a050fac,v1.1 +VMF00320-0005,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0005.fixmate.bam,e90bf36c17aa9ae9ed44b79e82235ccf,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0005.vcf.gz,123f019b74f83bf8626af367b0df2265,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0005.zarr.zip,dd57bacea1be68f813cb78dca315f6e5,v1.1 +VMF00320-0006,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0006.fixmate.bam,17dc5b2a9a2cdf82ee06d9cdc1153839,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0006.vcf.gz,dd6afc3932551463127365dcc017d142,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0006.zarr.zip,e0ce65e4ab1391e4d645975286533097,v1.1 +VMF00320-0007,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0007.fixmate.bam,a0fded1a777767539ee059a1ad39b788,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0007.vcf.gz,7962b1f97d0861ed28fc2005abf2198f,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0007.zarr.zip,e2a847b22979802af26a5777e927d186,v1.1 +VMF00320-0008,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0008.fixmate.bam,7d3bfd96eb7b0c0f47077eab269494de,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0008.vcf.gz,494992a8e86224a1fe2855b4818a4332,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0008.zarr.zip,fed98ea7b59f367d0524903ea27be46f,v1.1 +VMF00320-0009,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0009.fixmate.bam,a77f2de6aa947e426248718e1d997972,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0009.vcf.gz,566c9e5453d698596bb8b16fd7259a4e,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0009.zarr.zip,70eaa84840511e51eb67e4347bd920d2,v1.1 +VMF00320-0010,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0010.fixmate.bam,1c1512b9d3af504582467492f3bab54c,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0010.vcf.gz,a8aac7728c361a5b660ce71966793bee,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0010.zarr.zip,677ea2ff1ad9ca64c35a41a5096f556f,v1.1 +VMF00320-0012,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0012.fixmate.bam,45b301cd6c52f930df142ba9939555d6,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0012.vcf.gz,ad2eaa4f0e165592fdeb61da0b316dca,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0012.zarr.zip,deb70793b525b1d5bc8c5ce11a4723aa,v1.1 +VMF00320-0013,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0013.fixmate.bam,0297e8ad828f79f2149b1c309e577323,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0013.vcf.gz,9174904d1c8f320cd8a2f6d1630122bc,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0013.zarr.zip,07832d30432bc793f2967a84d879d793,v1.1 +VMF00320-0014,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0014.fixmate.bam,b5254f4e1e4a7e83dead193a552a7c03,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0014.vcf.gz,04beda86e879c0124dfabcd02957740e,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0014.zarr.zip,ab258044d10b1a9db8c1f9cb302d56ab,v1.1 +VMF00320-0015,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0015.fixmate.bam,4f8dca2f14095045e04753c416351c6a,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0015.vcf.gz,a56b674164bf73c3122991a9a32d0eb6,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0015.zarr.zip,d29816ca0f6a66772219c760a7564068,v1.1 +VMF00320-0016,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0016.fixmate.bam,1d3b63f99704ea1263c1b0d4f0756d36,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0016.vcf.gz,ae0df2238e513e5f6b36a2790602bc53,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0016.zarr.zip,15d3288af6196c8672345eb8d66455dd,v1.1 +VMF00320-0017,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0017.fixmate.bam,f9fd9e4bfba8a9cee00a722542ca66c1,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0017.vcf.gz,898992ae600bba20988f1d6d13fe36f1,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0017.zarr.zip,ca5ea83d03d963cf7d6b59fef2df8c3c,v1.1 +VMF00320-0018,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0018.fixmate.bam,19f76e7df9d64bd07f066036df0abe6a,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0018.vcf.gz,b11e77eadfa4ddbe6dd171e35ff4692b,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0018.zarr.zip,47408ed8594cdb3057f41a2f816ca173,v1.1 +VMF00320-0019,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0019.fixmate.bam,617580140c1998b2050cc7ecda3b5ccf,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0019.vcf.gz,03eadc4b084b0900425d50cf82e412ce,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0019.zarr.zip,85848c9d0af522579868076ec7d333e4,v1.1 +VMF00320-0020,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0020.fixmate.bam,bfd2526a9eb58889739ddac48f69b7f8,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0020.vcf.gz,06797c0837335dbcbdc2efa28c75d6c1,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0020.zarr.zip,63d79f9cd27da1df66d1a587bc1c63f7,v1.1 +VMF00320-0021,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0021.fixmate.bam,d11baeba898a101caca7ad49f3a46a6c,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0021.vcf.gz,3b51612224b89fb4ab81624b36519a1f,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0021.zarr.zip,e01815511e0271c52f4b435561413ac3,v1.1 +VMF00320-0022,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0022.fixmate.bam,d81f3539a217298762046973dc3b5978,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0022.vcf.gz,8a3baac22e21b4b48b2a0efb355f0573,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0022.zarr.zip,28e02cb22264cfae33ed52f883350774,v1.1 +VMF00320-0023,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0023.fixmate.bam,4e972747cbd7647e1e86bee1789166c3,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0023.vcf.gz,d33a75210b2fe81c76b7d7b5325d9fad,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0023.zarr.zip,ab05395b2c86bfbeb1eebb514488e639,v1.1 +VMF00320-0024,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0024.fixmate.bam,8467e2f1d168cd90c02d01c8b15a5dbf,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0024.vcf.gz,54c04501d321c14020624dcf79e17d8b,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0024.zarr.zip,a207d629dce5cc690d9dda886958e12f,v1.1 +VMF00320-0025,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0025.fixmate.bam,c0a63001d56b59734013653482c68387,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0025.vcf.gz,ea4f6b4b9c830f58c3c936bf145b2ec2,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0025.zarr.zip,ef69c07619d4ff5273b4ad30caf5ffaa,v1.1 +VMF00320-0026,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0026.fixmate.bam,f32db909fab4b51ca07d2e8a8d6cb3e3,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0026.vcf.gz,f620d83084ca4e7481e9ce10a5102347,https://1367-vo-af-donnelly-vmf00320-aste1.cog.sanger.ac.uk/VMF00320-0026.zarr.zip,4cc563b81039e91970e9e5950dfb02a8,v1.1 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/samples.meta.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/samples.meta.csv new file mode 100644 index 000000000..e155585a8 --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/samples.meta.csv @@ -0,0 +1,30 @@ +sample_id,partner_sample_id,contributor,country,location,year,month,latitude,longitude,sex_call +VMF00339-0001,KE83P4,Eric Ochomo,Kenya,Karare,2022,12,2.227,37.938,F +VMF00339-0002,KE9NX4,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,F +VMF00339-0003,KE9NX6,Eric Ochomo,Kenya,Kenya,2024,-1,1.442,38.431,F +VMF00339-0004,KE9NX7,Eric Ochomo,Kenya,Kenya,2024,-1,1.442,38.431,F +VMF00339-0005,KE9NXL,Eric Ochomo,Kenya,Kenya,2024,-1,1.442,38.431,M +VMF00339-0006,KE9NXM,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0007,KE9NXN,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0008,KE9NXO,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0009,KE9NXR,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0010,KE9NXS,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0011,KE9NXU,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0012,KE9NXV,Eric Ochomo,Kenya,Kenya,2024,-1,1.442,38.431,M +VMF00339-0013,KE9NXX,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0014,KE9NXY,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0015,KE9NXZ,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0016,KE9NY6,Eric Ochomo,Kenya,Kenya,2024,-1,1.442,38.431,F +VMF00339-0017,KE9NYA,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0018,KE9NYB,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,F +VMF00339-0019,KE9NYR,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,F +VMF00339-0020,KE9NYS,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,M +VMF00339-0021,KE9NYX,Eric Ochomo,Kenya,Kenya,2024,-1,1.442,38.431,F +VMF00339-0022,KE9NYZ,Eric Ochomo,Kenya,Mandera,2024,-1,3.935,41.855,F +VMF00339-0025,KE9OA4,Eric Ochomo,Kenya,Samburu,2024,-1,0.769,37.558,M +VMF00339-0026,KE9OA6,Eric Ochomo,Kenya,Samburu,2024,-1,0.769,37.558,M +VMF00339-0027,KE9OA7,Eric Ochomo,Kenya,Samburu,2024,-1,0.769,37.558,M +VMF00339-0028,KE9OA8,Eric Ochomo,Kenya,Samburu,2024,-1,0.769,37.558,F +VMF00339-0029,KE9OA9,Eric Ochomo,Kenya,Samburu,2024,-1,0.769,37.558,M +VMF00339-0030,KE9OAY,Eric Ochomo,Kenya,Samburu,2024,-1,0.769,37.558,M +VMF00339-0031,KE9OAZ,Eric Ochomo,Kenya,Samburu,2024,-1,0.769,37.558,M diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/sequence_qc_stats.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/sequence_qc_stats.csv new file mode 100644 index 000000000..a9cc7bc3c --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/sequence_qc_stats.csv @@ -0,0 +1,30 @@ +sample_id,mean_cov,median_cov,modal_cov,mean_cov_2RL,median_cov_2RL,mode_cov_2RL,mean_cov_3RL,median_cov_3RL,mode_cov_3RL,mean_cov_X,median_cov_X,mode_cov_X,frac_gen_cov,divergence,contam_pct,contam_LLR +VMF00339-0001,17.37,16,14,17.38,16,14,17.11,16,14,18.42,15,13,0.946,0.02138,0.0,0.0 +VMF00339-0002,24.61,24,24,24.74,24,24,24.37,24,24,24.97,24,23,0.953,0.01913,0.0,0.0 +VMF00339-0003,28.79,28,28,29.0,28,28,28.58,28,28,28.77,28,28,0.953,0.01901,0.0,0.0 +VMF00339-0004,28.2,28,27,28.38,28,28,28.04,28,27,28.08,27,27,0.952,0.01922,0.0,0.0 +VMF00339-0005,22.43,22,22,23.57,23,23,23.16,23,22,14.47,12,11,0.95,0.01924,0.0,0.0 +VMF00339-0006,17.63,17,17,18.55,18,17,18.2,18,17,11.34,9,8,0.951,0.01932,0.0,0.0 +VMF00339-0007,22.79,22,23,23.93,23,23,23.5,23,23,14.87,12,11,0.951,0.01928,0.0,0.0 +VMF00339-0008,21.8,21,22,22.94,22,22,22.43,22,22,14.26,11,10,0.952,0.01928,0.0,0.0 +VMF00339-0009,19.93,19,20,20.95,20,20,20.53,20,20,13.0,10,9,0.95,0.01927,0.0,0.0 +VMF00339-0010,22.39,22,22,23.55,23,22,23.04,23,22,14.63,12,11,0.951,0.01925,0.0,0.0 +VMF00339-0011,23.43,23,23,24.62,24,23,24.17,24,23,15.21,12,11,0.95,0.0193,0.0,0.0 +VMF00339-0012,25.73,25,26,27.09,26,26,26.67,26,26,16.0,13,12,0.951,0.01917,0.0,0.0 +VMF00339-0013,23.46,23,24,24.63,24,24,24.33,24,24,14.79,12,12,0.953,0.01913,0.0,0.0 +VMF00339-0014,22.16,22,22,23.32,23,22,22.87,22,22,14.22,12,11,0.951,0.01925,0.0,0.0 +VMF00339-0015,23.42,23,23,24.64,24,24,24.16,24,23,15.1,12,11,0.951,0.01925,0.0,0.0 +VMF00339-0016,22.75,22,22,22.88,22,22,22.53,22,22,23.08,21,20,0.955,0.01911,0.0,0.0 +VMF00339-0017,26.5,26,27,27.89,27,27,27.37,27,27,16.9,14,13,0.951,0.01925,0.0,0.0 +VMF00339-0018,22.36,22,22,22.53,22,22,22.13,22,22,22.61,21,21,0.952,0.01908,0.0,0.0 +VMF00339-0019,28.62,28,28,28.78,28,28,28.33,28,28,29.12,28,28,0.953,0.01914,0.0,0.0 +VMF00339-0020,22.78,22,23,24.03,23,23,23.5,23,23,14.35,12,11,0.951,0.01919,0.0,0.0 +VMF00339-0021,27.09,27,26,27.23,27,27,26.83,27,26,27.54,26,26,0.953,0.01919,0.0,0.0 +VMF00339-0022,24.24,24,24,24.39,24,24,23.94,24,24,24.86,24,24,0.952,0.01911,0.0,0.0 +VMF00339-0025,21.49,21,21,22.65,22,22,22.15,22,21,13.77,11,10,0.95,0.0193,0.0,0.0 +VMF00339-0026,24.5,24,25,25.8,25,25,25.22,25,25,15.97,13,12,0.951,0.01923,0.0,0.0 +VMF00339-0027,22.08,22,22,23.3,23,22,22.74,22,22,14.06,12,11,0.951,0.01931,0.0,0.0 +VMF00339-0028,29.04,29,28,29.18,29,28,28.81,29,28,29.42,28,28,0.952,0.01918,0.0,0.0 +VMF00339-0029,23.81,23,24,25.03,24,24,24.6,24,24,15.31,12,11,0.951,0.01933,0.0,0.0 +VMF00339-0030,23.05,23,23,24.24,23,23,23.78,23,23,14.87,12,11,0.951,0.01928,0.0,0.0 +VMF00339-0031,23.85,23,24,25.12,24,24,24.61,24,24,15.21,12,12,0.951,0.01928,0.0,0.0 diff --git a/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/wgs_snp_data.csv b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/wgs_snp_data.csv new file mode 100644 index 000000000..3d8e74b7e --- /dev/null +++ b/tests/anoph/fixture/vo_aste_release_master_us_central1/v1.0/metadata/general/1386-VO-KE-OCHOMO-VMF00339/wgs_snp_data.csv @@ -0,0 +1,30 @@ +sample_id,alignments_bam,alignments_bam_md5,snp_genotypes_vcf,snp_genotypes_vcf_md5,snp_genotypes_zarr,snp_genotypes_zarr_md5,pipeline_version +VMF00339-0001,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0001.fixmate.bam,ed596b25bf765009ccaf4d997d9bc7e7,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0001.vcf.gz,51e286c280ada7632ff6366628631fc4,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0001.zarr.zip,64aabc7ec5fe91c72f84ad375e285f74,v1.1 +VMF00339-0002,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0002.fixmate.bam,2d4336fbfc93bdbd1d871215dcf1371e,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0002.vcf.gz,15b7058aa0b3cddb594b4fee797b1dde,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0002.zarr.zip,547314e9f79405375c9b1e83826fba5a,v1.1 +VMF00339-0003,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0003.fixmate.bam,9f157609b17014440d527d8f3f91123e,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0003.vcf.gz,a83a3772719cbc8842978858075d8fb2,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0003.zarr.zip,29f5fa398647be4cef65cb9a352d4b6c,v1.1 +VMF00339-0004,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0004.fixmate.bam,e19fdd6ab82944d0c0382c493a270f57,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0004.vcf.gz,f9515d899a8b840a84fec168095df43b,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0004.zarr.zip,62b5895e26bf916e22f0713bd17a5a72,v1.1 +VMF00339-0005,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0005.fixmate.bam,839b93cd9f2f793d1dd46d66517e2315,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0005.vcf.gz,f5971308d2a3cbf8e817bf01f672bc05,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0005.zarr.zip,549a2eae34d7a441283e8995d4ddf63d,v1.1 +VMF00339-0006,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0006.fixmate.bam,61ab86e63db4fa8b4f7dc9d32bbb5ea9,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0006.vcf.gz,f41e872efd1051cf68a432cc59dfae72,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0006.zarr.zip,750571d515a4e89149072966d01ace57,v1.1 +VMF00339-0007,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0007.fixmate.bam,1bd3845484fcd8fbd24df11f3ab90cfd,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0007.vcf.gz,7844656e2cef9e9788ea9c5896165cc3,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0007.zarr.zip,c312e5f191b527f54e48298273f40ccc,v1.1 +VMF00339-0008,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0008.fixmate.bam,110a803e730a823add31b9cada965b3c,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0008.vcf.gz,e9ba0a81cdecfac30892baf31b3205ca,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0008.zarr.zip,1d9c080aec47ac7386ddafb5baae2889,v1.1 +VMF00339-0009,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0009.fixmate.bam,2097e55e57e67b7784c28766b946912d,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0009.vcf.gz,21dd5ae241278ec27baacbbe64fe61b1,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0009.zarr.zip,624faf386a94abb6aacd0a9151a535ee,v1.1 +VMF00339-0010,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0010.fixmate.bam,c5390c59e2d57da6a0632493a99b61d1,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0010.vcf.gz,dcca74768ba153de99bc40bc9fc8af87,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0010.zarr.zip,b4c23f28c5c35b4459c19ee91ce334bd,v1.1 +VMF00339-0011,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0011.fixmate.bam,736b2e2009d666362015ed8b0ddc824c,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0011.vcf.gz,e6d0af3368247bda990ea429dc464506,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0011.zarr.zip,6d77491b13ee97a7a73c786257073518,v1.1 +VMF00339-0012,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0012.fixmate.bam,7d9815c1b6648a8f0113a434b179e96d,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0012.vcf.gz,e500098a0e45f07ec818240901380e95,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0012.zarr.zip,eb2917cd171b93a4cd7f9c430fbb0c9b,v1.1 +VMF00339-0013,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0013.fixmate.bam,6878d03d9be954ce9ab030401239477e,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0013.vcf.gz,25f6fbcc4682cee362d9dc7361920ae4,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0013.zarr.zip,6310140e72f65a4320a7c9294ebb7048,v1.1 +VMF00339-0014,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0014.fixmate.bam,b6d2b385c17aa52851027b23ba665c96,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0014.vcf.gz,b4b2e18c7381c6a11ad23812c79e0038,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0014.zarr.zip,c62e2ded6d011a953348ddf7511aee08,v1.1 +VMF00339-0015,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0015.fixmate.bam,70cedb79314fb6ae9bba31efb904bcef,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0015.vcf.gz,9fe99ce6c192c9d92e68a9d90c9ec8ad,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0015.zarr.zip,d817dda5219ed0136a52979ba6dc478e,v1.1 +VMF00339-0016,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0016.fixmate.bam,1f5be6ca53b3ea1a9554718b6e10b312,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0016.vcf.gz,bf3cde2f9e3da6c4cf3ec30389d3f547,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0016.zarr.zip,0ec02e27d5fc280b1cf9f95d50db63fb,v1.1 +VMF00339-0017,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0017.fixmate.bam,bd985cd1585058d45780c72bb2f9c44e,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0017.vcf.gz,f290e30146da3f0e46a411ea39de283d,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0017.zarr.zip,dc7ea8f8f5d0b301bdace63a51e85ae2,v1.1 +VMF00339-0018,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0018.fixmate.bam,e3ac560a133668927441079091d225c4,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0018.vcf.gz,932940db3c4bdcf90e92f0244c11ab49,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0018.zarr.zip,4950c817d8002905a55818af5306ac3e,v1.1 +VMF00339-0019,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0019.fixmate.bam,2cf12209142d98e46e08286f5a5f1e16,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0019.vcf.gz,246a7e462b0c52530af13e6b89bea875,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0019.zarr.zip,2bcddbcab9cf693ff4e28e20648c5524,v1.1 +VMF00339-0020,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0020.fixmate.bam,1ef6cb6622be58687bc6227f21363695,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0020.vcf.gz,531c6efc850cc0be84f4aa60ca74ec6a,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0020.zarr.zip,99ace2b5fc8bec5cd711484a56203954,v1.1 +VMF00339-0021,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0021.fixmate.bam,1154d19eccb6d9294da227342aaf1271,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0021.vcf.gz,d3848f36c62367981a640821f5d354b7,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0021.zarr.zip,80ae03890f5a04db15982229579b9ce1,v1.1 +VMF00339-0022,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0022.fixmate.bam,16087be70f16ada61c5396d3ad937bb1,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0022.vcf.gz,b07e9df7d00d8cc51134205b84fb2175,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0022.zarr.zip,cc9cd5e883142060f7434436b4ca1514,v1.1 +VMF00339-0025,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0025.fixmate.bam,de2b6601faa84cf04bc384aaa859d36a,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0025.vcf.gz,33a670131dab16bc14a74a7c7c8154b5,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0025.zarr.zip,b658dbc770577070c6a320eb0292bae7,v1.1 +VMF00339-0026,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0026.fixmate.bam,e5f2b722c23294a86e92fda543d01ed5,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0026.vcf.gz,6bfd480e3c440159fec7ded908bea15b,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0026.zarr.zip,aacb99d388fb04dea5889895107474f1,v1.1 +VMF00339-0027,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0027.fixmate.bam,34c1d6c7ee914f029dbb198024627496,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0027.vcf.gz,2a413d7e5e1840881997dacef37da892,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0027.zarr.zip,a2e1335746cbe2aa4f81083d5afcd3af,v1.1 +VMF00339-0028,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0028.fixmate.bam,b7b2fcf8e4c4a20f69b20c5cc23fec6e,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0028.vcf.gz,5cfb108b1df2bbcd949d5a06cbf91183,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0028.zarr.zip,ea2cf549331fb3d9104fbe3fd72ab632,v1.1 +VMF00339-0029,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0029.fixmate.bam,83166f9dd74dd5dbe37b667e98248986,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0029.vcf.gz,f81a3e929dfa5af77aad522f5ed40c9d,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0029.zarr.zip,183e3ee9405acb4b3ed664eea01d2516,v1.1 +VMF00339-0030,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0030.fixmate.bam,8380cd85f19c6f0258a9048813600eee,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0030.vcf.gz,12b1c29cce15a53b4d24b677ab645ced,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0030.zarr.zip,dd0f22b2066a3150e8a6f3c59ca9d6c5,v1.1 +VMF00339-0031,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0031.fixmate.bam,1dbf7eead15efb03204b1ba9948c84d9,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0031.vcf.gz,a6dac1e16d8e2563c261c4f0efd53d4e,https://1386-vo-ke-ochomo-vmf00339-aste1.cog.sanger.ac.uk/VMF00339-0031.zarr.zip,ad97c8db866b3493149ac44a7014afd1,v1.1 diff --git a/tests/anoph/test_base.py b/tests/anoph/test_base.py index c49486a3f..181b36b3d 100644 --- a/tests/anoph/test_base.py +++ b/tests/anoph/test_base.py @@ -10,6 +10,8 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 +from malariagen_data import as1 as _as1 + from malariagen_data.anoph.base import AnophelesBase from malariagen_data.util import LoggingHelper @@ -62,6 +64,18 @@ def amin1_sim_api(amin1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesBase( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -90,6 +104,10 @@ def case_amin1_sim(amin1_sim_fixture, amin1_sim_api): return amin1_sim_fixture, amin1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @parametrize_with_cases("fixture,api", cases=".") def test_config(fixture, api): config = api.config diff --git a/tests/anoph/test_dipclust.py b/tests/anoph/test_dipclust.py index b355493e3..5a2d385e7 100644 --- a/tests/anoph/test_dipclust.py +++ b/tests/anoph/test_dipclust.py @@ -4,6 +4,8 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 +from malariagen_data import as1 as _as1 + from malariagen_data.anoph.dipclust import AnophelesDipClustAnalysis @@ -63,6 +65,24 @@ def af1_sim_api(af1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesDipClustAnalysis( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -83,6 +103,19 @@ def case_af1_sim(af1_sim_fixture, af1_sim_api): return af1_sim_fixture, af1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + +# Cases for tests that require CNV data (as1 fixture has no CNV data). +def case_cnv_ag3_sim(ag3_sim_fixture, ag3_sim_api): + return ag3_sim_fixture, ag3_sim_api + + +def case_cnv_af1_sim(af1_sim_fixture, af1_sim_api): + return af1_sim_fixture, af1_sim_api + + @pytest.mark.parametrize("sample_query", [None, "sex_call == 'F'"]) @pytest.mark.parametrize("distance_metric", ["cityblock", "euclidean"]) @parametrize_with_cases("fixture,api", cases=".") @@ -201,7 +234,7 @@ def test_plot_diplotype_clustering_advanced_with_transcript( @pytest.mark.parametrize("sample_query", [None, "sex_call == 'F'"]) -@parametrize_with_cases("fixture,api", cases=".") +@parametrize_with_cases("fixture,api", cases=".", prefix="case_cnv_") def test_plot_diplotype_clustering_advanced_with_cnv_region( fixture, api: AnophelesDipClustAnalysis, sample_query ): diff --git a/tests/anoph/test_distance.py b/tests/anoph/test_distance.py index 44a24e52b..420e8f2cc 100644 --- a/tests/anoph/test_distance.py +++ b/tests/anoph/test_distance.py @@ -6,6 +6,8 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 +from malariagen_data import as1 as _as1 + from malariagen_data.anoph.distance import AnophelesDistanceAnalysis from malariagen_data.anoph import pca_params @@ -73,6 +75,24 @@ def adir1_sim_api(adir1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesDistanceAnalysis( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -97,6 +117,10 @@ def case_adir1_sim(adir1_sim_fixture, adir1_sim_api): return adir1_sim_fixture, adir1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + def check_biallelic_diplotype_pairwise_distance(*, api, data_params, metric): # Check available data. ds = api.biallelic_snp_calls(**data_params) diff --git a/tests/anoph/test_fst.py b/tests/anoph/test_fst.py index 813485003..35727ac78 100644 --- a/tests/anoph/test_fst.py +++ b/tests/anoph/test_fst.py @@ -9,6 +9,8 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 +from malariagen_data import as1 as _as1 + from malariagen_data.anoph.fst import AnophelesFstAnalysis @@ -76,6 +78,24 @@ def adir1_sim_api(adir1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesFstAnalysis( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -100,6 +120,10 @@ def case_adir1_sim(adir1_sim_fixture, adir1_sim_api): return adir1_sim_fixture, adir1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @parametrize_with_cases("fixture,api", cases=".") def test_fst_gwss(fixture, api: AnophelesFstAnalysis): # Set up test parameters. diff --git a/tests/anoph/test_g123.py b/tests/anoph/test_g123.py index 80ca3a9e3..4c152ce1c 100644 --- a/tests/anoph/test_g123.py +++ b/tests/anoph/test_g123.py @@ -7,6 +7,8 @@ from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 from malariagen_data import amin1 as _amin1 +from malariagen_data import as1 as _as1 + from malariagen_data.anoph.g123 import AnophelesG123Analysis @@ -95,6 +97,25 @@ def amin1_sim_api(amin1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesG123Analysis( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + default_phasing_analysis=None, + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -123,6 +144,10 @@ def case_amin1_sim(amin1_sim_fixture, amin1_sim_api): return amin1_sim_fixture, amin1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + def check_g123_gwss(*, api, g123_params): # Run main gwss function under test. x, g123 = api.g123_gwss(**g123_params) diff --git a/tests/anoph/test_genome_features.py b/tests/anoph/test_genome_features.py index ba7563d97..50c048a62 100644 --- a/tests/anoph/test_genome_features.py +++ b/tests/anoph/test_genome_features.py @@ -7,7 +7,8 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 -from malariagen_data import adir1 as _adir1 +from malariagen_data import as1 as _as1 + from malariagen_data.anoph.genome_features import AnophelesGenomeFeaturesData from malariagen_data.util import Region, _resolve_region @@ -44,13 +45,13 @@ def af1_sim_api(af1_sim_fixture): @pytest.fixture -def adir1_sim_api(adir1_sim_fixture): +def as1_sim_api(as1_sim_fixture): return AnophelesGenomeFeaturesData( - url=adir1_sim_fixture.url, - public_url=adir1_sim_fixture.url, - config_path=_adir1.CONFIG_PATH, - major_version_number=_adir1.MAJOR_VERSION_NUMBER, - major_version_path=_adir1.MAJOR_VERSION_PATH, + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, pre=False, gff_gene_type="protein_coding_gene", gff_gene_name_attribute="Note", @@ -66,8 +67,8 @@ def case_af1_sim(af1_sim_fixture, af1_sim_api): return af1_sim_fixture, af1_sim_api -def case_adir1_sim(adir1_sim_fixture, adir1_sim_api): - return adir1_sim_fixture, adir1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api gff3_cols = [ @@ -111,10 +112,10 @@ def test_genome_features_default_attributes_af1( assert df_gf.columns.to_list() == expected_cols -def test_genome_features_default_attributes_adir1( - adir1_sim_api: AnophelesGenomeFeaturesData, +def test_genome_features_default_attributes_as1( + as1_sim_api: AnophelesGenomeFeaturesData, ): - df_gf = adir1_sim_api.genome_features() + df_gf = as1_sim_api.genome_features() assert isinstance(df_gf, pd.DataFrame) expected_cols = gff3_cols + ["ID", "Parent", "Note", "description"] assert df_gf.columns.to_list() == expected_cols diff --git a/tests/anoph/test_genome_sequence.py b/tests/anoph/test_genome_sequence.py index 8550341b2..a50dcfaac 100644 --- a/tests/anoph/test_genome_sequence.py +++ b/tests/anoph/test_genome_sequence.py @@ -8,6 +8,8 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 +from malariagen_data import as1 as _as1 + from malariagen_data.anoph.genome_sequence import AnophelesGenomeSequenceData from malariagen_data.util import Region @@ -37,6 +39,18 @@ def af1_sim_api(af1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesGenomeSequenceData( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + ) + + def case_ag3_sim(ag3_sim_fixture, ag3_sim_api): return ag3_sim_fixture, ag3_sim_api @@ -45,6 +59,10 @@ def case_af1_sim(af1_sim_fixture, af1_sim_api): return af1_sim_fixture, af1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @parametrize_with_cases("fixture,api", cases=".") def test_contigs(fixture, api): contigs = api.contigs diff --git a/tests/anoph/test_heterozygosity.py b/tests/anoph/test_heterozygosity.py index 10118bbf3..45d42a7d0 100644 --- a/tests/anoph/test_heterozygosity.py +++ b/tests/anoph/test_heterozygosity.py @@ -8,6 +8,8 @@ from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 from malariagen_data import amin1 as _amin1 +from malariagen_data import as1 as _as1 + from malariagen_data.anoph.heterozygosity import AnophelesHetAnalysis @@ -92,6 +94,24 @@ def amin1_sim_api(amin1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesHetAnalysis( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + ) + + def case_ag3_sim(ag3_sim_fixture, ag3_sim_api): return ag3_sim_fixture, ag3_sim_api @@ -108,6 +128,10 @@ def case_amin1_sim(amin1_sim_fixture, amin1_sim_api): return amin1_sim_fixture, amin1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @parametrize_with_cases("fixture,api", cases=".") def test_plot_heterozygosity_track(fixture, api: AnophelesHetAnalysis): # Set up test parameters. diff --git a/tests/anoph/test_ld.py b/tests/anoph/test_ld.py index 96eb9ec91..601b47509 100644 --- a/tests/anoph/test_ld.py +++ b/tests/anoph/test_ld.py @@ -5,6 +5,7 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 +from malariagen_data import as1 as _as1 from malariagen_data.anoph.ld import AnophelesLdAnalysis @@ -54,6 +55,24 @@ def af1_sim_api(af1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesLdAnalysis( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + ) + + def case_ag3_sim(ag3_sim_fixture, ag3_sim_api): return ag3_sim_fixture, ag3_sim_api @@ -62,6 +81,10 @@ def case_af1_sim(af1_sim_fixture, af1_sim_api): return af1_sim_fixture, af1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @parametrize_with_cases("fixture,api", cases=".") def test_ld_pruning_returns_fewer_snps(fixture, api: AnophelesLdAnalysis): region = random.choice(api.contigs) diff --git a/tests/anoph/test_pca.py b/tests/anoph/test_pca.py index 3516f9db7..c2964bdb4 100644 --- a/tests/anoph/test_pca.py +++ b/tests/anoph/test_pca.py @@ -8,6 +8,7 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 +from malariagen_data import as1 as _as1 from malariagen_data.anoph.pca import AnophelesPca from malariagen_data.anoph import pca_params @@ -76,6 +77,24 @@ def adir1_sim_api(adir1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesPca( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -100,6 +119,10 @@ def case_adir1_sim(adir1_sim_fixture, adir1_sim_api): return adir1_sim_fixture, adir1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @parametrize_with_cases("fixture,api", cases=".") def test_pca_plotting(fixture, api: AnophelesPca): # Parameters for selecting input data. diff --git a/tests/anoph/test_plink_converter.py b/tests/anoph/test_plink_converter.py index d926525ce..e3533bca8 100644 --- a/tests/anoph/test_plink_converter.py +++ b/tests/anoph/test_plink_converter.py @@ -5,6 +5,7 @@ from malariagen_data import af1 as _af1 from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 +from malariagen_data import as1 as _as1 from malariagen_data.anoph.to_plink import PlinkConverter @@ -77,6 +78,24 @@ def adir1_sim_api(adir1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return PlinkConverter( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -97,6 +116,10 @@ def case_af1_sim(af1_sim_fixture, af1_sim_api): return af1_sim_fixture, af1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @parametrize_with_cases("fixture,api", cases=".") def test_plink_converter(fixture, api: PlinkConverter, tmp_path): # Parameters for selecting input data, filtering, and converting. diff --git a/tests/anoph/test_sample_metadata.py b/tests/anoph/test_sample_metadata.py index 59211924a..12e74e99a 100644 --- a/tests/anoph/test_sample_metadata.py +++ b/tests/anoph/test_sample_metadata.py @@ -14,6 +14,7 @@ from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 from malariagen_data import amin1 as _amin1 +from malariagen_data import as1 as _as1 from malariagen_data.anoph.sample_metadata import AnophelesSampleMetadata @@ -192,6 +193,19 @@ def af1_sim_unrestricted_surveillance_use_only_api(af1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesSampleMetadata( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + taxon_colors=_as1.TAXON_COLORS, + ) + + @pytest.fixture def missing_metadata_api(fixture_dir): # In this fixture, one of the sample sets (AG1000G-BF-A) has missing files @@ -234,6 +248,11 @@ def case_amin1_sim(amin1_sim_fixture, amin1_sim_api): return amin1_sim_fixture, amin1_sim_api +@case(tags="as1") +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @case def case_ag3_sim_unrestricted_use_only( ag3_sim_fixture, ag3_sim_unrestricted_use_only_api @@ -1097,7 +1116,7 @@ def test_wgs_data_catalog(fixture, api): assert set(df["sample_id"]) == set(df_samples["sample_id"]) -@parametrize_with_cases("fixture,api", cases=".") +@parametrize_with_cases("fixture,api", cases=".", filter=~ft.has_tag("as1")) def test_wgs_run_accessions(fixture, api): # Set up test. df_sample_sets = api.sample_sets().set_index("sample_set") diff --git a/tests/anoph/test_snp_data.py b/tests/anoph/test_snp_data.py index f8540ba94..bcaf58aa8 100644 --- a/tests/anoph/test_snp_data.py +++ b/tests/anoph/test_snp_data.py @@ -16,6 +16,7 @@ from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 from malariagen_data import amin1 as _amin1 +from malariagen_data import as1 as _as1 from malariagen_data.anoph.base_params import DEFAULT @@ -99,6 +100,23 @@ def amin1_sim_api(amin1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesSnpData( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -131,6 +149,11 @@ def case_amin1_sim(amin1_sim_fixture, amin1_sim_api): return amin1_sim_fixture, amin1_sim_api +@case +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @parametrize_with_cases("fixture,api", cases=".") def test_open_snp_sites(fixture, api: AnophelesSnpData): root = api.open_snp_sites() diff --git a/tests/anoph/test_snp_frq.py b/tests/anoph/test_snp_frq.py index 1e5bda529..4736ccff5 100644 --- a/tests/anoph/test_snp_frq.py +++ b/tests/anoph/test_snp_frq.py @@ -11,6 +11,7 @@ from malariagen_data import ag3 as _ag3 from malariagen_data import adir1 as _adir1 from malariagen_data import amin1 as _amin1 +from malariagen_data import as1 as _as1 from malariagen_data.anoph.snp_frq import AnophelesSnpFrequencyAnalysis @@ -105,6 +106,24 @@ def amin1_sim_api(amin1_sim_fixture): ) +@pytest.fixture +def as1_sim_api(as1_sim_fixture): + return AnophelesSnpFrequencyAnalysis( + url=as1_sim_fixture.url, + public_url=as1_sim_fixture.url, + config_path=_as1.CONFIG_PATH, + major_version_number=_as1.MAJOR_VERSION_NUMBER, + major_version_path=_as1.MAJOR_VERSION_PATH, + pre=False, + gff_gene_type="protein_coding_gene", + gff_gene_name_attribute="Note", + gff_default_attributes=("ID", "Parent", "Note", "description"), + default_site_mask="stephensi", + results_cache=as1_sim_fixture.results_cache_path.as_posix(), + taxon_colors=_as1.TAXON_COLORS, + ) + + # N.B., here we use pytest_cases to parametrize tests. Each # function whose name begins with "case_" defines a set of # inputs to the test functions. See the documentation for @@ -129,6 +148,10 @@ def case_adir1_sim(adir1_sim_fixture, adir1_sim_api): return adir1_sim_fixture, adir1_sim_api +def case_as1_sim(as1_sim_fixture, as1_sim_api): + return as1_sim_fixture, as1_sim_api + + @case(tags="single-sampleset") def case_amin1_sim(amin1_sim_fixture, amin1_sim_api): return amin1_sim_fixture, amin1_sim_api