Merge branch 'master' into cache-bugs

Author: jonbrenas

malariagen_data/anoph/fst.py

@@ -363,7 +363,7 @@ def average_fst(
         region: base_params.region,
         cohort1_query: base_params.sample_query,
         cohort2_query: base_params.sample_query,
-        sample_query_options: Optional[base_params.sample_query] = None,
+        sample_query_options: Optional[base_params.sample_query_options] = None,
         sample_sets: Optional[base_params.sample_sets] = None,
         cohort_size: Optional[base_params.cohort_size] = fst_params.cohort_size_default,
         min_cohort_size: Optional[

malariagen_data/anoph/hap_data.py

@@ -64,7 +64,7 @@ def _prep_phasing_analysis_param(self, *, analysis: hap_params.analysis) -> str:
             return analysis
         else:
             raise ValueError(
-                f"Invalid phasing analysis, must be one of f{self.phasing_analysis_ids}."
+                f"Invalid phasing analysis, must be one of {self.phasing_analysis_ids}."
             )
 
     @_check_types

malariagen_data/anoph/snp_data.py

@@ -116,9 +116,7 @@ def _prep_site_mask_param(
         elif site_mask in self.site_mask_ids:
             return site_mask
         else:
-            raise ValueError(
-                f"Invalid site mask, must be one of f{self.site_mask_ids}."
-            )
+            raise ValueError(f"Invalid site mask, must be one of {self.site_mask_ids}.")
 
     def _prep_optional_site_mask_param(
         self,

malariagen_data/veff.py

@@ -548,7 +548,21 @@ def _get_within_intron_effect(base_effect, intron):
             effect = base_effect._replace(effect="INTRONIC", impact="MODIFIER")
 
     else:
-        # TODO intronic INDELs and MNPs
-        effect = base_effect._replace(effect="TODO intronic indels and MNPs")
+        # INDELs and MNPs — use the closest edge of the variant to the splice site
+        if strand == "+":
+            dist_5prime = ref_start - (intron_start - 1)
+            dist_3prime = -(ref_stop - (intron_stop + 1))
+        else:
+            dist_5prime = (intron_stop + 1) - ref_stop
+            dist_3prime = -((intron_start - 1) - ref_start)
+
+        indel_min_dist = min(dist_5prime, dist_3prime)
+
+        if indel_min_dist <= 2:
+            effect = base_effect._replace(effect="SPLICE_CORE", impact="HIGH")
+        elif indel_min_dist <= 7:
+            effect = base_effect._replace(effect="SPLICE_REGION", impact="MODERATE")
+        else:
+            effect = base_effect._replace(effect="INTRONIC", impact="MODIFIER")
 
     return effect