Correct breaking of Scaffolds on Haplotigs or Unlocs

sanger-tol · Nov 15, 2024 · 602a31d · 602a31d
1 parent 2f68fba
commit 602a31d
Show file tree

Hide file tree

Showing 8 changed files with 26 additions and 22 deletions.
diff --git a/src/tola/assembly/build_assembly.py b/src/tola/assembly/build_assembly.py
@@ -279,25 +279,24 @@ def assemblies_with_scaffolds_fused(self) -> list[Assembly]:
 
     def scaffolds_fused_by_name(self) -> Iterator[Scaffold]:
         gap = self.default_gap
-        new_scffld = None
-        current_hap_chr = None, None
+        hap_name_scaffold = {}
         for scffld in self.scaffolds:
             if not scffld.rows:
                 # discard_overhanging_fragments() may have removed the only
                 # row from an OverlapResult
                 continue
-            hap_chr = scffld.haplotype, scffld.name
-            if hap_chr != current_hap_chr:
-                if new_scffld:
-                    yield new_scffld
-                current_hap_chr = hap_chr
-                new_scffld = Scaffold(
-                    scffld.name, tag=scffld.tag, haplotype=scffld.haplotype
-                )
+            build_scffld = hap_name_scaffold.setdefault(
+                (scffld.haplotype, scffld.name),
+                Scaffold(
+                    scffld.name,
+                    tag=scffld.tag,
+                    haplotype=scffld.haplotype,
+                ),
+            )
             if isinstance(scffld, OverlapResult):
-                new_scffld.append_scaffold(scffld.to_scaffold(), gap)
+                build_scffld.append_scaffold(scffld.to_scaffold(), gap)
             else:
-                new_scffld.append_scaffold(scffld)
+                build_scffld.append_scaffold(scffld)
 
-        if new_scffld:
-            yield new_scffld
+        for scffld in hap_name_scaffold.values():
+            yield scffld
diff --git a/tests/data/ilIthSala1_1/ilIthSala1_1-pretext-to-tpf.log b/tests/data/ilIthSala1_1/ilIthSala1_1-pretext-to-tpf.log
@@ -201,7 +201,7 @@ ilIthSala1_1-pretext-to-tpf_HAP1
             651,870  HAP1_SCAFFOLD_54
                 ...  ...
              16,408  HAP1_SCAFFOLD_864
-         92,470,266  bp total
+         92,524,843  bp total
 
 ilIthSala1_1-pretext-to-tpf_HAP2
         389,828,879  bp sequence (minus gaps)
@@ -223,4 +223,4 @@ ilIthSala1_1-pretext-to-tpf_Haplotigs
     n = 1
             309,695  H_1
 
-Curation made 37 cuts in contigs, 87 breaks at gaps and 185 joins
+Curation made 37 cuts in contigs, 87 breaks at gaps and 186 joins
diff --git a/tests/data/ilIthSala1_1/ilIthSala1_1-pretext-to-tpf_HAP1.tpf b/tests/data/ilIthSala1_1/ilIthSala1_1-pretext-to-tpf_HAP1.tpf
@@ -305,6 +305,7 @@ GAP	TYPE-2	200
 GAP	TYPE-2	200
 ?	HAP1_SCAFFOLD_1:18881119-21883956	Z	PLUS
 ?	HAP1_SCAFFOLD_48:409325-463901	HAP1_SCAFFOLD_48	MINUS
+GAP	TYPE-2	200
 ?	HAP1_SCAFFOLD_48:1-136441	HAP1_SCAFFOLD_48	PLUS
 ?	HAP1_SCAFFOLD_54:1-651870	HAP1_SCAFFOLD_54	PLUS
 ?	HAP1_SCAFFOLD_55:1-597161	HAP1_SCAFFOLD_55	PLUS

diff --git a/tests/data/ilIthSala1_1/ilIthSala1_1-pretext-to-tpf_info.yaml b/tests/data/ilIthSala1_1/ilIthSala1_1-pretext-to-tpf_info.yaml
@@ -1,10 +1,10 @@
 assemblies:
   HAP1:
     manual_breaks: 32
-    manual_joins: 97
+    manual_joins: 98
   HAP2:
     manual_breaks: 55
     manual_joins: 88
 manual_breaks: 87
-manual_joins: 185
+manual_joins: 186
 manual_haplotig_removals: 1
diff --git a/tests/data/ngHelPoly1_1/chrs.csv b/tests/data/ngHelPoly1_1/chrs.csv
@@ -1,6 +1,6 @@
 RL_1,RL_1_unloc_1,RL_1_unloc_2,RL_1_unloc_3,RL_1_unloc_4,RL_1_unloc_5,RL_1_unloc_6,RL_1_unloc_7,RL_1_unloc_8,RL_1_unloc_9,RL_1_unloc_10,RL_1_unloc_11,RL_1_unloc_12,RL_1_unloc_13
-RL_2,RL_2,RL_2,RL_2,RL_2_unloc_1,RL_2_unloc_2,RL_2_unloc_3,RL_2_unloc_4,RL_2_unloc_5,RL_2_unloc_6,RL_2_unloc_7,RL_2_unloc_8,RL_2_unloc_9,RL_2_unloc_10,RL_2_unloc_11,RL_2_unloc_12
+RL_2,RL_2_unloc_1,RL_2_unloc_2,RL_2_unloc_3,RL_2_unloc_4,RL_2_unloc_5,RL_2_unloc_6,RL_2_unloc_7,RL_2_unloc_8,RL_2_unloc_9,RL_2_unloc_10,RL_2_unloc_11,RL_2_unloc_12
 RL_3,RL_3_unloc_1,RL_3_unloc_2,RL_3_unloc_3,RL_3_unloc_4,RL_3_unloc_5,RL_3_unloc_6,RL_3_unloc_7,RL_3_unloc_8,RL_3_unloc_9,RL_3_unloc_10
 RL_4,RL_4_unloc_1,RL_4_unloc_2,RL_4_unloc_3,RL_4_unloc_4,RL_4_unloc_5,RL_4_unloc_6,RL_4_unloc_7
-RL_5,RL_5,RL_5_unloc_1,RL_5_unloc_2,RL_5_unloc_3,RL_5_unloc_4,RL_5_unloc_5,RL_5_unloc_6,RL_5_unloc_7,RL_5_unloc_8,RL_5_unloc_9,RL_5_unloc_10,RL_5_unloc_11
+RL_5,RL_5_unloc_1,RL_5_unloc_2,RL_5_unloc_3,RL_5_unloc_4,RL_5_unloc_5,RL_5_unloc_6,RL_5_unloc_7,RL_5_unloc_8,RL_5_unloc_9,RL_5_unloc_10,RL_5_unloc_11
 X,X_unloc_1,X_unloc_2,X_unloc_3,X_unloc_4,X_unloc_5,X_unloc_6,X_unloc_7,X_unloc_8,X_unloc_9
diff --git a/tests/data/ngHelPoly1_1/ngHelPoly1_1-pretext-to-tpf.log b/tests/data/ngHelPoly1_1/ngHelPoly1_1-pretext-to-tpf.log
@@ -81,4 +81,4 @@ ngHelPoly1_1-pretext-to-tpf_Haplotigs
              30,757  H_24
           2,164,462  bp total
 
-Curation made 9 cuts in contigs, 210 breaks at gaps and 468 joins
+Curation made 9 cuts in contigs, 210 breaks at gaps and 472 joins
diff --git a/tests/data/ngHelPoly1_1/ngHelPoly1_1-pretext-to-tpf.tpf b/tests/data/ngHelPoly1_1/ngHelPoly1_1-pretext-to-tpf.tpf
@@ -1413,6 +1413,7 @@ GAP	TYPE-2	200
 ?	scaffold_2:22958123-23757084	RL_2	PLUS
 GAP	TYPE-2	200
 ?	scaffold_2:23757285-23952476	RL_2	PLUS
+GAP	TYPE-2	200
 ?	scaffold_2:24023605-24245085	RL_2	PLUS
 GAP	TYPE-2	200
 ?	scaffold_2:24245286-24337062	RL_2	PLUS
@@ -2286,6 +2287,7 @@ GAP	TYPE-2	200
 ?	scaffold_2:84005701-84076804	RL_2	PLUS
 GAP	TYPE-2	200
 ?	scaffold_2:84077005-84270004	RL_2	PLUS
+GAP	TYPE-2	200
 ?	scaffold_2:84324151-84517061	RL_2	PLUS
 GAP	TYPE-2	200
 ?	scaffold_2:84517262-84628943	RL_2	PLUS
@@ -2363,6 +2365,7 @@ GAP	TYPE-2	200
 ?	scaffold_2:93382642-93420642	RL_2	PLUS
 GAP	TYPE-2	200
 ?	scaffold_2:93420843-93689450	RL_2	PLUS
+GAP	TYPE-2	200
 ?	scaffold_2:93772333-93813810	RL_2	PLUS
 GAP	TYPE-2	200
 ?	scaffold_2:93814011-94409008	RL_2	PLUS
@@ -5070,6 +5073,7 @@ GAP	TYPE-2	200
 ?	scaffold_378:1-30343	RL_5	PLUS
 GAP	TYPE-2	200
 ?	scaffold_5:42746124-43081660	RL_5	PLUS
+GAP	TYPE-2	200
 ?	scaffold_5:43175773-43394772	RL_5	PLUS
 GAP	TYPE-2	200
 ?	scaffold_471:1-24981	RL_5	PLUS

diff --git a/tests/data/ngHelPoly1_1/ngHelPoly1_1-pretext-to-tpf_info.yaml b/tests/data/ngHelPoly1_1/ngHelPoly1_1-pretext-to-tpf_info.yaml
@@ -1,5 +1,5 @@
 assemblies:
   Primary:
     manual_breaks: 210
-    manual_joins: 468
+    manual_joins: 472
 manual_haplotig_removals: 24