refact(mi): move common overlap checking code to function

lint(mi): remove debug log
davidlougheed · Jul 11, 2024 · 4e25a17 · 4e25a17
1 parent 21014f3
commit 4e25a17
Show file tree

Hide file tree

Showing 6 changed files with 15 additions and 27 deletions.
diff --git a/strkit/mi/base.py b/strkit/mi/base.py
@@ -89,6 +89,14 @@ def get_loci_overlapping(self, contig: str, start: int, end: int) -> list[tuple[
     def should_exclude_locus(self, contig: str, start: int, end: int) -> bool:
         return any(True for _ in overlapping_loci_dict_of_list(contig, start, end, self._exclude_dict))
 
+    def should_skip_locus(self, contig: str, start: int, end: int) -> bool:
+        # Check to make sure call is present in TRF BED file, if it is specified
+        # Check to make sure the locus is not excluded via overlap with exclude BED
+        return (
+            (self._loci_file and self._loci_dict and not self.get_loci_overlapping(contig, start, end))
+            or self.should_exclude_locus(contig, start, end)
+        )
+
     @abstractmethod
     def _get_sample_contigs(self, include_sex_chromosomes: bool = False) -> tuple[set, set, set]:
         return set(), set(), set()

diff --git a/strkit/mi/expansionhunter.py b/strkit/mi/expansionhunter.py
@@ -52,10 +52,7 @@ def calculate_contig(self, contig: str) -> MIContigResult:
 
             k = (contig, cv.start, cv.stop)
 
-            # Check to make sure call is present in TRF BED file, if it is specified
-            # Check to make sure the locus is not excluded via overlap with exclude BED
-            if ((self._loci_file and self._loci_dict and not self.get_loci_overlapping(*k))
-                    or self.should_exclude_locus(*k)):
+            if self.should_skip_locus(*k):
                 continue
 
             cr.seen_locus(*k)

diff --git a/strkit/mi/gangstr.py b/strkit/mi/gangstr.py
@@ -36,11 +36,8 @@ def calculate_contig(self, contig: str) -> MIContigResult:
             # Check to make sure call is present in TRF BED file, if it is specified
             k1 = (contig, cv.start, cv.stop)
             k2 = (contig, cv.start + 1, cv.stop + 1)
-            if (self._loci_file and self._loci_dict and not self.get_loci_overlapping(*k1)
-                    and not self.get_loci_overlapping(*k2)):
-                continue
 
-            if self.should_exclude_locus(*k1) or self.should_exclude_locus(*k2):
+            if self.should_skip_locus(*k1) or self.should_skip_locus(*k2):
                 continue
 
             cr.seen_locus(*k1)

diff --git a/strkit/mi/repeathmm.py b/strkit/mi/repeathmm.py
@@ -55,10 +55,7 @@ def calculate_contig(self, contig: str) -> MIContigResult:
                 k = (contig, locus_start, locus_end)
 
                 # Check to make sure call is present in TRF BED file, if it is specified
-                if self._loci_file and self._loci_dict and not self.get_loci_overlapping(*k):
-                    continue
-
-                if self.should_exclude_locus(*k):
+                if self.should_skip_locus(*k):
                     continue
 
                 cr.seen_locus(*k)

diff --git a/strkit/mi/strkit.py b/strkit/mi/strkit.py
@@ -72,8 +72,7 @@ def calculate_contig(self, contig: str) -> MIContigResult:
                 start = int(locus_data[1])
                 end = int(locus_data[2])
 
-                # Check to make sure call is present in TRF BED file, if it is specified
-                if self._loci_file and self._loci_dict and not self.get_loci_overlapping(contig, start, end):
+                if self.should_skip_locus(contig, start, end):
                     continue
 
                 # Check to make sure call is present in all trio individuals
@@ -203,10 +202,7 @@ def calculate_contig(self, contig: str) -> MIContigResult:
             k = (contig, int(locus_start), int(locus_end))
 
             # Check to make sure call is present in TRF BED file, if it is specified
-            if self._loci_file and self._loci_dict and not self.get_loci_overlapping(*k):
-                continue
-
-            if self.should_exclude_locus(*k):
+            if self.should_skip_locus(*k):
                 continue
 
             cr.seen_locus(*k)
@@ -282,14 +278,12 @@ def calculate_contig(self, contig: str) -> MIContigResult:
             # TODO: Handle sex chromosomes
 
             # Check to make sure call is present in TRF BED file, if it is specified
-            if ((self._loci_file and self._loci_dict and not self.get_loci_overlapping(*k))
-                    or self.should_exclude_locus(*k)):
+            if self.should_skip_locus(*k):
                 continue
 
             cr.seen_locus(*k)
 
             if mv is None or fv is None:
-                print("!!", k, list(mvf.fetch(*k)))
                 # Variant isn't found in at least one of the parents, so we can't do anything with it.
                 # TODO: We need to actually check calls, and check with sample ID, not just assume
                 continue

diff --git a/strkit/mi/tandem_genotypes.py b/strkit/mi/tandem_genotypes.py
@@ -51,12 +51,7 @@ def calculate_contig(self, contig: str) -> MIContigResult:
 
                 k = (contig, int(lookup[1]), int(lookup[2]))
 
-                # Check to make sure call is present in TRF BED file, if it is specified
-                if self._loci_file and self._loci_dict and not self.get_loci_overlapping(*k):
-                    continue
-
-                # noinspection PyTypeChecker
-                if self.should_exclude_locus(*k):
+                if self.should_skip_locus(*k):
                     continue
 
                 cr.seen_locus(*k)