De-underscored some functions

Author: jonbrenas

malariagen_data/anoph/hap_frq.py

@@ -9,9 +9,9 @@
 from ..util import (
     check_types,
     haplotype_frequencies,
-    _prep_samples_for_cohort_grouping,
-    _build_cohorts_from_sample_grouping,
-    _add_frequency_ci,
+    prep_samples_for_cohort_grouping,
+    build_cohorts_from_sample_grouping,
+    add_frequency_ci,
 )
 from .hap_data import AnophelesHapData
 from .sample_metadata import locate_cohorts
@@ -161,7 +161,7 @@ def haplotypes_frequencies_advanced(
         )
 
         # Prepare sample metadata for cohort grouping.
-        df_samples = _prep_samples_for_cohort_grouping(
+        df_samples = prep_samples_for_cohort_grouping(
             df_samples=df_samples,
             area_by=area_by,
             period_by=period_by,
@@ -171,7 +171,7 @@ def haplotypes_frequencies_advanced(
         group_samples_by_cohort = df_samples.groupby(["taxon", "area", "period"])
 
         # Build cohorts dataframe.
-        df_cohorts = _build_cohorts_from_sample_grouping(
+        df_cohorts = build_cohorts_from_sample_grouping(
             group_samples_by_cohort=group_samples_by_cohort,
             min_cohort_size=min_cohort_size,
         )
@@ -272,6 +272,6 @@ def haplotypes_frequencies_advanced(
         )
 
         # Add confidence intervals.
-        _add_frequency_ci(ds=ds_out, ci_method=ci_method)
+        add_frequency_ci(ds=ds_out, ci_method=ci_method)
 
         return ds_out

malariagen_data/anoph/snp_frq.py

@@ -14,9 +14,9 @@
 from ..util import (
     check_types,
     pandas_apply,
-    _prep_samples_for_cohort_grouping,
-    _build_cohorts_from_sample_grouping,
-    _add_frequency_ci,
+    prep_samples_for_cohort_grouping,
+    build_cohorts_from_sample_grouping,
+    add_frequency_ci,
 )
 from .snp_data import AnophelesSnpData
 from .sample_metadata import locate_cohorts
@@ -439,7 +439,7 @@ def snp_allele_frequencies_advanced(
         )
 
         # Prepare sample metadata for cohort grouping.
-        df_samples = _prep_samples_for_cohort_grouping(
+        df_samples = prep_samples_for_cohort_grouping(
             df_samples=df_samples,
             area_by=area_by,
             period_by=period_by,
@@ -449,7 +449,7 @@ def snp_allele_frequencies_advanced(
         group_samples_by_cohort = df_samples.groupby(["taxon", "area", "period"])
 
         # Build cohorts dataframe.
-        df_cohorts = _build_cohorts_from_sample_grouping(
+        df_cohorts = build_cohorts_from_sample_grouping(
             group_samples_by_cohort=group_samples_by_cohort,
             min_cohort_size=min_cohort_size,
         )
@@ -603,7 +603,7 @@ def snp_allele_frequencies_advanced(
             ds_out = ds_out.isel(variants=loc_variants)
 
         # Add confidence intervals.
-        _add_frequency_ci(ds=ds_out, ci_method=ci_method)
+        add_frequency_ci(ds=ds_out, ci_method=ci_method)
 
         # Tidy up display by sorting variables.
         sorted_vars: List[str] = sorted([str(k) for k in ds_out.keys()])
@@ -739,7 +739,7 @@ def aa_allele_frequencies_advanced(
             ds_aa_frq = ds_aa_frq.isel(variants=loc_variants)
 
         # Compute new confidence intervals.
-        _add_frequency_ci(ds=ds_aa_frq, ci_method=ci_method)
+        add_frequency_ci(ds=ds_aa_frq, ci_method=ci_method)
 
         # Tidy up display by sorting variables.
         ds_aa_frq = ds_aa_frq[sorted(ds_aa_frq)]

malariagen_data/anopheles.py

@@ -60,9 +60,9 @@
     region_str,
     simple_xarray_concat,
     pandas_apply,
-    _add_frequency_ci,
-    _build_cohorts_from_sample_grouping,
-    _prep_samples_for_cohort_grouping,
+    prep_samples_for_cohort_grouping,
+    build_cohorts_from_sample_grouping,
+    add_frequency_ci,
 )
 
 DEFAULT_MAX_COVERAGE_VARIANCE = 0.2
@@ -1359,7 +1359,7 @@ def _gene_cnv_frequencies_advanced(
         df_samples = df_samples.set_index("sample_id").loc[sample_id].reset_index()
 
         debug("prepare sample metadata for cohort grouping")
-        df_samples = _prep_samples_for_cohort_grouping(
+        df_samples = prep_samples_for_cohort_grouping(
             df_samples=df_samples,
             area_by=area_by,
             period_by=period_by,
@@ -1369,7 +1369,7 @@ def _gene_cnv_frequencies_advanced(
         group_samples_by_cohort = df_samples.groupby(["taxon", "area", "period"])
 
         debug("build cohorts dataframe")
-        df_cohorts = _build_cohorts_from_sample_grouping(
+        df_cohorts = build_cohorts_from_sample_grouping(
             group_samples_by_cohort=group_samples_by_cohort,
             min_cohort_size=min_cohort_size,
         )
@@ -1474,7 +1474,7 @@ def _gene_cnv_frequencies_advanced(
             ds_out = ds_out.isel(variants=loc_variants)
 
         debug("add confidence intervals")
-        _add_frequency_ci(ds=ds_out, ci_method=ci_method)
+        add_frequency_ci(ds=ds_out, ci_method=ci_method)
 
         debug("tidy up display by sorting variables")
         ds_out = ds_out[sorted(ds_out)]

malariagen_data/util.py

@@ -1587,7 +1587,7 @@ def _karyotype_tags_n_alt(gt, alts, inversion_alts):
     return inv_n_alt
 
 
-def _prep_samples_for_cohort_grouping(*, df_samples, area_by, period_by):
+def prep_samples_for_cohort_grouping(*, df_samples, area_by, period_by):
     # Take a copy, as we will modify the dataframe.
     df_samples = df_samples.copy()
 
@@ -1622,7 +1622,7 @@ def _prep_samples_for_cohort_grouping(*, df_samples, area_by, period_by):
     return df_samples
 
 
-def _build_cohorts_from_sample_grouping(*, group_samples_by_cohort, min_cohort_size):
+def build_cohorts_from_sample_grouping(*, group_samples_by_cohort, min_cohort_size):
     # Build cohorts dataframe.
     df_cohorts = group_samples_by_cohort.agg(
         size=("sample_id", len),
@@ -1659,7 +1659,7 @@ def _build_cohorts_from_sample_grouping(*, group_samples_by_cohort, min_cohort_s
     return df_cohorts
 
 
-def _add_frequency_ci(*, ds, ci_method):
+def add_frequency_ci(*, ds, ci_method):
     from statsmodels.stats.proportion import proportion_confint  # type: ignore
 
     if ci_method is not None: