Merge branch 'develop' into remove-zfpkm

Author: JoshLoecker

main/como/cluster_rnaseq.py

@@ -8,7 +8,7 @@
 import numpy as np
 
 from como.data_types import LogLevel
-from como.utils import _log_and_raise_error, stringlist_to_list
+from como.utils import log_and_raise_error, stringlist_to_list
 
 
 @dataclass
@@ -35,35 +35,35 @@ def __post_init__(self):  # noqa: C901, ignore too complex
             self.seed = np.random.randint(0, 100_000)
 
         if (isdigit(self.min_active_count) and int(self.min_active_count) < 0) or self.min_active_count != "default":
-            _log_and_raise_error(
+            log_and_raise_error(
                 "min_active_count must be either 'default' or an integer > 0",
                 error=ValueError,
                 level=LogLevel.ERROR,
             )
 
         if (isdigit(self.quantile) and 0 > int(self.quantile) > 100) or self.quantile != "default":
-            _log_and_raise_error(
+            log_and_raise_error(
                 "quantile must be either 'default' or an integer between 0 and 100",
                 error=ValueError,
                 level=LogLevel.ERROR,
             )
 
         if (isdigit(self.replicate_ratio) and 0 > self.replicate_ratio > 1.0) or self.replicate_ratio != "default":
-            _log_and_raise_error(
+            log_and_raise_error(
                 "--rep-ratio must be either 'default' or a float between 0 and 1",
                 error=ValueError,
                 level=LogLevel.ERROR,
             )
 
         if (isdigit(self.batch_ratio) and 0 > self.batch_ratio > 1.0) or self.batch_ratio != "default":
-            _log_and_raise_error(
+            log_and_raise_error(
                 "--batch-ratio must be either 'default' or a float between 0 and 1",
                 error=ValueError,
                 level=LogLevel.ERROR,
             )
 
         if self.filtering_technique.lower() not in {"quantile", "tpm", "cpm", "zfpkm"}:
-            _log_and_raise_error(
+            log_and_raise_error(
                 "--technique must be either 'quantile', 'tpm', 'cpm', 'zfpkm'",
                 error=ValueError,
                 level=LogLevel.ERROR,
@@ -73,35 +73,35 @@ def __post_init__(self):  # noqa: C901, ignore too complex
             self.filtering_technique = "quantile"
 
         if self.cluster_algorithm.lower() not in {"mca", "umap"}:
-            _log_and_raise_error(
+            log_and_raise_error(
                 "--clust_algo must be either 'mca', 'umap'",
                 error=ValueError,
                 level=LogLevel.ERROR,
             )
 
         if 0 > self.min_distance > 1.0:
-            _log_and_raise_error(
+            log_and_raise_error(
                 "--min_dist must be a float between 0 and 1",
                 error=ValueError,
                 level=LogLevel.ERROR,
             )
 
         if (isdigit(self.num_replicate_neighbors) and self.num_replicate_neighbors < 1) or self.num_replicate_neighbors != "default":
-            _log_and_raise_error(
+            log_and_raise_error(
                 "--n-neighbors-rep must be either 'default' or an integer > 1",
                 error=ValueError,
                 level=LogLevel.ERROR,
             )
 
         if (isdigit(self.num_batch_neighbors) and self.num_batch_neighbors < 1) or self.num_batch_neighbors != "default":
-            _log_and_raise_error(
+            log_and_raise_error(
                 "--n-neighbors-batch must be either 'default' or an integer > 1",
                 error=ValueError,
                 level=LogLevel.ERROR,
             )
 
         if (isdigit(self.num_context_neighbors) and self.num_context_neighbors < 1) or self.num_context_neighbors != "default":
-            _log_and_raise_error(
+            log_and_raise_error(
                 "--n-neighbors-context must be either 'default' or an integer > 1",
                 error=ValueError,
                 level=LogLevel.ERROR,

main/como/combine_distributions.py

@@ -18,7 +18,7 @@
     _OutputCombinedSourceFilepath,
     _SourceWeights,
 )
-from como.utils import LogLevel, _log_and_raise_error, _num_columns
+from como.utils import LogLevel, log_and_raise_error, _num_columns, get_missing_gene_data
 
 
 def _combine_z_distribution_for_batch(
@@ -186,7 +186,7 @@ def _combine_z_distribution_for_context(
     for res in zscore_results:
         matrix = res.z_score_matrix.copy()
         if len(matrix.columns) > 1:
-            _log_and_raise_error(
+            log_and_raise_error(
                 f"Expected a single column for combined z-score dataframe for data '{res.type.value.lower()}'. Got '{len(matrix.columns)}' columns",
                 error=ValueError,
                 level=LogLevel.ERROR,
@@ -302,7 +302,7 @@ async def _begin_combining_distributions(
             else ""
         )
         if not index_name:
-            _log_and_raise_error(
+            log_and_raise_error(
                 f"Unable to find common gene identifier across batches for source '{source.value}' in context '{context_name}'",
                 error=ValueError,
                 level=LogLevel.ERROR,