Added nonredundant command to find non-redundant list of centromeres …

…from AS-HOR array lengths. Refactored integration function to allow multiple expected outputs to check.

.gitattributes

@@ -13,3 +13,12 @@ test/status/input/chr9_cens_partials.fa.out filter=lfs diff=lfs merge=lfs -text
 test/status/input/chm13_chm1_cens_v21.trimmed.fa.noheader.out filter=lfs diff=lfs merge=lfs -text
 test/status/input/chm13_chm1_cens_v21.trimmed.fa.noheader.out.png filter=lfs diff=lfs merge=lfs -text
 test/status/input/chr21_cens.png filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/expected/unaccounted_left.tsv filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/expected/uniq_left.tsv filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/expected/uniq_right.tsv filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/input/hifiasm_all_AS-HOR_lengths.tsv filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/expected/both.tsv filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/expected/dupes_right.tsv filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/input/verkko_all_AS-HOR_lengths.tsv filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/expected/dupes_left.tsv filter=lfs diff=lfs merge=lfs -text
+test/nonredundant/expected/unaccounted_right.tsv filter=lfs diff=lfs merge=lfs -text

censtats/main.py

@@ -3,6 +3,7 @@
 
 from .status.cli import add_status_cli, check_cens_status
 from .length.cli import add_hor_length_cli, calculate_hor_length
+from .nonredundant.cli import add_nonredundant_cli, get_nonredundant_cens
 
 if TYPE_CHECKING:
     SubArgumentParser = argparse._SubParsersAction[argparse.ArgumentParser]
@@ -15,6 +16,7 @@ def main() -> int:
     sub_ap = ap.add_subparsers(dest="cmd")
     add_status_cli(sub_ap)
     add_hor_length_cli(sub_ap)
+    add_nonredundant_cli(sub_ap)
 
     args = ap.parse_args()
 
@@ -35,6 +37,18 @@ def main() -> int:
         return calculate_hor_length(
             args.input, args.bp_jump_thr, args.arr_len_thr, args.output
         )
+    elif args.cmd == "nonredundant":
+        return get_nonredundant_cens(
+            args.infile_left,
+            args.infile_right,
+            args.outfile_left,
+            args.outfile_right,
+            args.outfile_both,
+            args.unaccounted_left,
+            args.unaccounted_right,
+            args.duplicates_left,
+            args.duplicates_right,
+        )
     else:
         raise ValueError(f"Unknown command: {args.cmd}")