Merge branch 'master' into GH1151-fix-roh-hmm-cache-name

Author: khushthecoder

malariagen_data/adar1.py

@@ -130,7 +130,6 @@ def __init__(
             tqdm_class=tqdm_class,
             taxon_colors=TAXON_COLORS,
             virtual_contigs=None,
-            gene_names=None,
             inversion_tag_path=None,
             unrestricted_use_only=unrestricted_use_only,
             surveillance_use_only=surveillance_use_only,

malariagen_data/adir1.py

@@ -130,7 +130,6 @@ def __init__(
             tqdm_class=tqdm_class,
             taxon_colors=TAXON_COLORS,
             virtual_contigs=None,
-            gene_names=None,
             inversion_tag_path=None,
             unrestricted_use_only=unrestricted_use_only,
             surveillance_use_only=surveillance_use_only,

malariagen_data/af1.py

@@ -132,7 +132,6 @@ def __init__(
             tqdm_class=tqdm_class,
             taxon_colors=TAXON_COLORS,
             virtual_contigs=None,
-            gene_names=None,
             inversion_tag_path=None,
             unrestricted_use_only=unrestricted_use_only,
             surveillance_use_only=surveillance_use_only,

malariagen_data/ag3.py

@@ -212,7 +212,6 @@ def __init__(
             taxon_colors=TAXON_COLORS,
             aim_species_colors=AIM_SPECIES_COLORS,
             virtual_contigs=VIRTUAL_CONTIGS,
-            gene_names=GENE_NAMES,
             inversion_tag_path=INVERSION_TAG_PATH,
             unrestricted_use_only=unrestricted_use_only,
             surveillance_use_only=surveillance_use_only,

malariagen_data/amin1.py

@@ -130,7 +130,6 @@ def __init__(
             tqdm_class=tqdm_class,
             taxon_colors=TAXON_COLORS,
             virtual_contigs=None,
-            gene_names=None,
             inversion_tag_path=None,
             unrestricted_use_only=unrestricted_use_only,
             surveillance_use_only=surveillance_use_only,

malariagen_data/anoph/aim_data.py

@@ -40,10 +40,24 @@ def __init__(
         # to the superclass constructor.
         super().__init__(**kwargs)
 
-        # Store possible values for the `aims` parameter.
-        # TODO Consider moving this to data resource configuration.
-        self._aim_ids = aim_ids
-        self._aim_palettes = aim_palettes
+        # Read AIM parameters from the JSON config, falling back to
+        # constructor args for backward compatibility.
+        config = self.config
+        _aim_ids = config.get("AIM_IDS", None)
+        if _aim_ids is not None:
+            self._aim_ids: Optional[aim_params.aim_ids] = tuple(_aim_ids)
+        else:
+            self._aim_ids = aim_ids
+
+        _aim_palettes = config.get("AIM_PALETTES", None)
+        if _aim_palettes is not None:
+            # Convert lists to tuples for each palette entry.
+            self._aim_palettes: Optional[aim_params.aim_palettes] = {
+                k: tuple(v)
+                for k, v in _aim_palettes.items()  # type: ignore
+            }
+        else:
+            self._aim_palettes = aim_palettes
 
         # Set up caches.
         self._cache_aim_variants: Dict[str, xr.Dataset] = dict()

malariagen_data/anoph/genome_features.py

@@ -23,9 +23,9 @@ class AnophelesGenomeFeaturesData(AnophelesGenomeSequenceData):
     def __init__(
         self,
         *,
-        gff_gene_type: str,
-        gff_gene_name_attribute: str,
-        gff_default_attributes: Tuple[str, ...],
+        gff_gene_type: Optional[str] = None,
+        gff_gene_name_attribute: Optional[str] = None,
+        gff_default_attributes: Optional[Tuple[str, ...]] = None,
         gene_names: Optional[Mapping[str, str]] = None,
         **kwargs,
     ):
@@ -34,16 +34,30 @@ def __init__(
         # to the superclass constructor.
         super().__init__(**kwargs)
 
-        # TODO Consider moving these parameters to configuration, as they could
-        # change if the GFF ever changed.
-        self._gff_gene_type = gff_gene_type
-        self._gff_gene_name_attribute = gff_gene_name_attribute
-        self._gff_default_attributes = gff_default_attributes
+        # Read GFF parameters from the JSON config, falling back to
+        # constructor args for backward compatibility.
+        config = self.config
+        self._gff_gene_type = config.get("GFF_GENE_TYPE", gff_gene_type or "gene")
+        self._gff_gene_name_attribute = config.get(
+            "GFF_GENE_NAME_ATTRIBUTE", gff_gene_name_attribute or "Name"
+        )
+        _default_attrs = config.get("GFF_DEFAULT_ATTRIBUTES", None)
+        if _default_attrs is not None:
+            self._gff_default_attributes = tuple(_default_attrs)
+        elif gff_default_attributes is not None:
+            self._gff_default_attributes = gff_default_attributes
+        else:
+            self._gff_default_attributes = ("ID", "Parent", "Name", "description")
 
         # Allow manual override of gene names.
-        if gene_names is None:
-            gene_names = dict()
-        self._gene_name_overrides = gene_names
+        # Read from config if available, falling back to constructor arg.
+        _gene_names = config.get("GENE_NAMES", None)
+        if _gene_names is not None:
+            self._gene_name_overrides = _gene_names
+        elif gene_names is not None:
+            self._gene_name_overrides = gene_names
+        else:
+            self._gene_name_overrides = dict()
 
         # Setup caches.
         self._cache_genome_features: Dict[Tuple[str, ...], pd.DataFrame] = dict()

malariagen_data/anopheles.py

@@ -136,7 +136,6 @@ def __init__(
         taxon_colors: Optional[Mapping[str, str]] = None,
         aim_species_colors: Optional[Mapping[str, str]] = None,
         virtual_contigs: Optional[Mapping[str, Sequence[str]]] = None,
-        gene_names: Optional[Mapping[str, str]] = None,
         inversion_tag_path: Optional[str] = None,
         unrestricted_use_only: Optional[bool] = None,
         surveillance_use_only: Optional[bool] = None,
@@ -174,7 +173,6 @@ def __init__(
             taxon_colors=taxon_colors,
             aim_species_colors=aim_species_colors,
             virtual_contigs=virtual_contigs,
-            gene_names=gene_names,
             inversion_tag_path=inversion_tag_path,
             unrestricted_use_only=unrestricted_use_only,
             surveillance_use_only=surveillance_use_only,

malariagen_data/util.py

@@ -361,7 +361,7 @@ def _dask_compress_dataarray(a, indexer, indexer_computed, dim):
 
 
 def _da_compress(
-    indexer: da.Array | np.ndarray,
+    indexer: Union[da.Array, np.ndarray],
     data: da.Array,
     axis: int,
     indexer_computed: Optional[np.ndarray] = None,