add io test

BALSAMIC/assets/scripts/postprocess_gens_cnvkit.py

@@ -1,8 +1,5 @@
 import click
 from BALSAMIC.utils.io import read_csv
-import numpy as np
-import warnings
-
 
 def calculate_log2_ratio(purity, log2_ratio, ploidy):
     # Ensure that the inputs are within valid ranges

BALSAMIC/commands/config/case.py

@@ -139,14 +139,15 @@ def case_config(
             gnomad_min_af5=gnomad_min_af5,
             panel_bed=panel_bed,
         )
-        # Update references dictionary with GENS reference-files
-        references.update(
-            {
-                gens_file: path
-                for gens_file, path in gens_references.items()
-                if path is not None
-            }
-        )
+        if gens_references:
+            # Update references dictionary with GENS reference-files
+            references.update(
+                {
+                    gens_file: path
+                    for gens_file, path in gens_references.items()
+                    if path is not None
+                }
+            )
 
     variants_observations = {
         "clinical_snv_observations": clinical_snv_observations,

BALSAMIC/utils/cli.py

@@ -220,18 +220,18 @@ def get_gens_references(
     gens_coverage_pon: Optional[str],
     gnomad_min_af5: Optional[str],
     panel_bed: Optional[str],
-) -> Dict[str, str]:
+) -> Dict[str, str] | None:
     """
     Assigns reference-files required for GENS if they have been supplied, else exists with error message.
 
     Args:
-        genome_interval: Optional[str]. Coverage-regions. (required for WGS GENS)
-        gens_coverage_pon: Optional[str]. PON for GATK CollectReadCounts. (required for WGS GENS)
+        genome_interval: Optional[str] Coverage-regions. (required for WGS GENS)
+        gens_coverage_pon: Optional[str] PON for GATK CollectReadCounts. (required for WGS GENS)
         gnomad_min_af5: Optional[str] gnomad VCF filtered to keep variants above 5% VAF (required for WGS and TGA GENS).
         panel_bed: Optional[str] Bedfile supplied for TGA analyses.
 
     Returns:
-         Dict[str, str] with paths to GENS reference files
+         Dict[str, str] with paths to GENS reference files or None
     """
 
     if panel_bed and gnomad_min_af5:
@@ -243,14 +243,11 @@ def get_gens_references(
             "gens_coverage_pon": gens_coverage_pon,
             "gnomad_min_af5": gnomad_min_af5,
         }
-
-    error_message = (
-        "GENS reference file is always required to run BALSAMIC.\n"
-        "WGS requires arguments: genome_interval, gens_coverage_pon, gnomad_min_af5\n"
-        "TGA requires argument: gnomad_min_af5"
-    )
-    LOG.error(error_message)
-    raise BalsamicError(error_message)
+    if any([gnomad_min_af5, genome_interval, gens_coverage_pon]):
+        error_message = "GENS for WGS requires all arguments: genome_interval, gens_coverage_pon, gnomad_min_af5"
+        LOG.error(error_message)
+        raise BalsamicError(error_message)
+    return None
 
 
 def get_bioinfo_tools_version(

tests/commands/config/test_config_sample.py

@@ -420,12 +420,12 @@ def test_missing_required_gens_arguments_wgs(
     # THEN the CLI should exit code 2 and display an informative error message
     assert result.exit_code == 1
     assert (
-        "WGS requires arguments: genome_interval, gens_coverage_pon, gnomad_min_af5"
+        "GENS for WGS requires all arguments: genome_interval, gens_coverage_pon, gnomad_min_af5"
         in result.output
     )
 
 
-def test_missing_required_gens_arguments_tga(
+def test_missing_gens_arguments_tga(
     invoke_cli,
     tumor_sample_name: str,
     analysis_dir: str,
@@ -461,9 +461,8 @@ def test_missing_required_gens_arguments_tga(
             sentieon_license,
         ],
     )
-    # THEN the CLI should exit code 2 and display an informative error message
-    assert result.exit_code == 1
-    assert "TGA requires argument: gnomad_min_af5" in result.output
+    # THEN the CLI should exit code 0
+    assert result.exit_code == 0
 
 
 def test_config_with_gens_arguments(

tests/utils/test_utils.py

@@ -46,6 +46,7 @@
 from BALSAMIC.utils.exc import BalsamicError, WorkflowRunError
 from BALSAMIC.utils.io import (
     read_json,
+    read_csv,
     read_vcf_file,
     read_yaml,
     write_finish_file,
@@ -479,6 +480,20 @@ def test_read_json(config_path: str):
     assert type(config_dict) is dict
 
 
+def test_read_csv(purity_csv_path: str):
+    """Test data extraction from a CSV file."""
+
+    # GIVEN a CSV path
+
+    # WHEN calling the function
+    csv_list: list[Dict] = read_csv(purity_csv_path)
+
+    # THEN the config.json file should be correctly parsed
+    assert len(csv_list) == 1
+    assert csv_list[0]["Purity"] == "0.64"
+    assert csv_list[0]["Sampleid"] == "tumor.initial"
+
+
 def test_read_json_error():
     """Test data extraction from a BALSAMIC config JSON file for an invalid path."""