Fixes #240: convert alt case/control labels to Unknown

src/cgr_gwas_qc/reporting/qc_exclusions.py

@@ -40,6 +40,15 @@ def sample_exclusion_counts(data: DataFrameOrFile) -> pd.DataFrame:
 
     return _create_sample_exclusion_counts_table(df)
 
+def _replace_unknown_labels(df):
+	known_cc_labels = ["Unknown", "Case", "Control", "QC"] # must be one of these labels
+
+	# Use boolean indexing to select rows where "case_control" isn't in known_cc_labels
+	is_unknown = ~df["case_control"].isin(known_cc_labels)
+	# Update the "case_control" column with "Unknown" for these rows
+	df.loc[is_unknown, "case_control"] = "Unknown"
+	return df
+
 
 def _create_sample_exclusion_counts_table(df: pd.DataFrame) -> pd.DataFrame:
     row_names = {
@@ -62,6 +71,8 @@ def _create_sample_exclusion_counts_table(df: pd.DataFrame) -> pd.DataFrame:
         "total": "All Samples",  # added below
     }
 
+    df = _replace_unknown_labels(df.copy())
+
     return (
         df.assign(is_array_processing_failure=lambda x: x.is_missing_idats | x.is_missing_gtc)
         .assign(samples_remaining=lambda x: ~x.analytic_exclusion & ~x.is_internal_control)
@@ -77,6 +88,7 @@ def _create_sample_exclusion_counts_table(df: pd.DataFrame) -> pd.DataFrame:
         .groupby("case_control")
         .sum()  # Count the number True
         .pipe(_clean_up_counts_table, row_names=row_names, col_names=col_names)
+
     )
 
 
@@ -113,6 +125,8 @@ def _create_subject_exclusion_counts_table(df: pd.DataFrame) -> pd.DataFrame:
         "total": "All Subjects",
     }
 
+    df = _replace_unknown_labels(df.copy())
+
     return (
         df.reindex(["case_control", *row_names.keys()], axis=1)
         .groupby("case_control")