From cfc80b6ec8e2f36e0e338eb26a38b1072c113dcc Mon Sep 17 00:00:00 2001
From: David Lougheed <david.lougheed@gmail.com>
Date: Tue, 12 Nov 2024 15:53:50 -0500
Subject: [PATCH] chore(mi): LongTR: report skip reason

---
 strkit/mi/base.py   | 27 +++++++++++++++++----------
 strkit/mi/longtr.py |  4 ++--
 2 files changed, 19 insertions(+), 12 deletions(-)

diff --git a/strkit/mi/base.py b/strkit/mi/base.py
index 7b927cb..3805bf4 100644
--- a/strkit/mi/base.py
+++ b/strkit/mi/base.py
@@ -107,23 +107,30 @@ def get_loci_overlapping(
         self, contig: str, start: int, end: int, first_only: bool
     ) -> list[tuple[int, int, list[str]]]:
         return overlapping_loci_dict_of_dict(
-            contig, start, end, self._loci_dict,  first_only, dict_cache_key=self._loci_dict_cache_key
+            contig, start, end, self._loci_dict, first_only, dict_cache_key=self._loci_dict_cache_key
         )
 
     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, True))
 
-    def should_skip_locus(self, contig: str, start: int, end: int, cached_overlapping: Optional[list] = None) -> bool:
+    def should_skip_locus(
+        self, contig: str, start: int, end: int, cached_overlapping: Optional[list] = None
+    ) -> str | None:
+        # Returns either a reason string (if yes) or None (=== no)
+
         # 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 (cached_overlapping or self.get_loci_overlapping(contig, start, end, True))
-            )
-            or self.should_exclude_locus(contig, start, end)
-        )
+
+        if not self._loci_file or not self._loci_dict:
+            return None
+
+        if not (cached_overlapping or self.get_loci_overlapping(contig, start, end, True)):
+            return "no overlapping loci"
+
+        if self.should_exclude_locus(contig, start, end):
+            return "should_exclude_locus returned True"
+
+        return None
 
     @abstractmethod
     def _get_sample_contigs(self) -> tuple[set, set, set]:
diff --git a/strkit/mi/longtr.py b/strkit/mi/longtr.py
index d17adb8..9cd55cc 100644
--- a/strkit/mi/longtr.py
+++ b/strkit/mi/longtr.py
@@ -37,8 +37,8 @@ def calculate_contig(self, contig: str) -> MIContigResult:
 
             overlapping = self.get_loci_overlapping(k[0], k[1], k[2], True)
 
-            if self.should_skip_locus(k[0], k[1], k[2], cached_overlapping=overlapping):
-                self._logger.debug(f"Skipping locus {k}: no overlap with BED")
+            if r := self.should_skip_locus(k[0], k[1], k[2], cached_overlapping=overlapping):
+                self._logger.debug(f"Skipping locus {k}: {r}")
                 continue
 
             cr.seen_locus(*k)