Handle two-part contig names again and put them under test

vgteam · Aug 4, 2023 · 92d4953 · 92d4953 · adamnovak · Aug 5, 2023
1 parent 40d98fb
commit 92d4953
Show file tree

Hide file tree

Showing 4 changed files with 56 additions and 3 deletions.
diff --git a/deps/libhandlegraph b/deps/libhandlegraph
diff --git a/src/unittest/handle.cpp b/src/unittest/handle.cpp
@@ -15,6 +15,7 @@
 #include "bdsg/hash_graph.hpp"
 
 #include <handlegraph/util.hpp>
+#include <handlegraph/path_metadata.hpp>
 
 #include <iostream>
 #include <limits>
@@ -2542,5 +2543,32 @@ TEST_CASE("handlegraph PathMetadata name format preserves ranges on generic path
     REQUIRE(subrange.second == PathMetadata::NO_END_POSITION);
 }
 
+TEST_CASE("handlegraph PathMetadata name format can parse two-part names", "[handle]") {
+    std::string path_name = "GRCh38#chr1";
+
+    PathSense sense;
+    string sample;
+    string locus;
+    size_t haplotype;
+    size_t phase_block;
+    subrange_t subrange;
+    PathMetadata::parse_path_name(path_name,
+                                  sense,
+                                  sample,
+                                  locus,
+                                  haplotype,
+                                  phase_block,
+                                  subrange);
+
+    REQUIRE(PathMetadata::parse_sample_name(path_name) == sample);
+    REQUIRE(PathMetadata::parse_locus_name(path_name) == locus);
+
+    REQUIRE(sense == PathSense::REFERENCE); 
+    REQUIRE(sample == "GRCh38");
+    REQUIRE(locus == "chr1");
+    REQUIRE(phase_block == PathMetadata::NO_PHASE_BLOCK);
+    REQUIRE(subrange == PathMetadata::NO_SUBRANGE);
+}
+
 }
 }
diff --git a/test/graphs/gfa_two_part_reference.gfa b/test/graphs/gfa_two_part_reference.gfa
@@ -0,0 +1,20 @@
+H	VN:Z:1.1	RS:Z:GRCh37 GRCh38
+S	1	G
+S	2	A
+S	4	GGG
+S	5	T
+S	6	A
+S	7	C
+S	8	A
+S	9	A
+L	1	+	2	+	0M
+L	1	+	4	+	0M
+L	2	+	4	+	0M
+L	4	+	5	+	0M
+L	5	+	6	+	0M
+L	6	+	7	+	0M
+L	6	+	8	+	0M
+L	7	+	9	+	0M
+L	8	+	9	+	0M
+P	GRCh38#chr1	1+,4+,5+,6+,7+,9+	*,*,*,*,*
+P	GRCh37#chr1	1+,2+,4+,5+,6+,8+,9+	*,*,*,*,*,*
diff --git a/test/t/48_vg_convert.t b/test/t/48_vg_convert.t
@@ -7,7 +7,7 @@ PATH=../bin:$PATH # for vg
 
 export LC_ALL="C" # force a consistent sort order
 
-plan tests 100
+plan tests 102
 
 vg construct -r complex/c.fa -v complex/c.vcf.gz > c.vg
 cat <(vg view c.vg | grep ^S | sort) <(vg view c.vg | grep L | uniq | wc -l) <(vg paths -v c.vg -E) > c.info
@@ -411,8 +411,13 @@ vg convert -a graphs/components_paths_rgfa.gfa > components_paths_rgfa.hg
 is "${?}" "0" "GFA -> HashGraph conversion works with redundant paths"
 is "$(vg paths --list -x components_paths_rgfa.hg | wc -l)" "1" "GFA -> HashGraph conversion with redundant paths keeps one copy of the redundant path"
 
+# We should be able to handle pseudo-PanSN paths where there is no haplotype
+vg convert -a graphs/gfa_two_part_reference.gfa > gfa_two_part_reference.hg
+is "${?}" "0" "GFA -> HashGraph conversion works with two-part reference path names"
+is "$(vg paths -M -x gfa_two_part_reference.hg | grep REFERENCE | wc -l)" "2" "GFA -> HashGraph conversion with with two-part reference path names gets the right paths"
+
 rm -f paths.truth.txt paths.gbz.txt paths.gfa.txt paths.hg.txt
-rm -f gfa_with_reference.gbz rgfa_with_reference.gbz gfa_with_reference.hg components_paths_rgfa.hg rgfa_with_reference.hg extracted.gfa 
+rm -f gfa_with_reference.gbz rgfa_with_reference.gbz gfa_with_reference.hg components_paths_rgfa.hg gfa_two_part_reference.hg rgfa_with_reference.hg extracted.gfa 
 
 #####
 # GFA Streaming