Fix reference allele frequency calculation

PathoGenOmics-Lab · Oct 3, 2023 · e48cc62 · e48cc62
1 parent e5d77c1
commit e48cc62
Showing 1 changed file with 6 additions and 2 deletions.
diff --git a/workflow/scripts/weighted_distances.py b/workflow/scripts/weighted_distances.py
@@ -57,8 +57,12 @@ def get_frequencies_in_position(site_df: pd.DataFrame, pos: int, reference: Seq)
     for alt in site_df["ALT"]:
         frequencies[alt] = float(site_df.loc[site_df["ALT"] == alt, "ALT_FREQ"].iloc[0])
     # Obtain frequency for reference allele
-    ref = 1 - sum(frequencies.values())
-    frequencies[reference[pos-1]] += ref
+    reference_frequency = 1 - sum(frequencies.values())
+    reference_allele = reference[pos-1]
+    if reference_allele in frequencies.keys():
+        frequencies[reference_allele] += reference_frequency
+    else:
+        frequencies[reference_allele] = reference_frequency
     return tuple(frequencies[alt] for alt in site_df["ALT"].unique())