Fix off by one bug when converting BAM files to d4 files. (#97)

* Fix off by one bug when converting BAM files to d4 files.

Fix off by one bug when converting BAM files to d4 files.

Fixes: #75

* First iteration adding d4tools create tests

* Make sure the tests are in order

---------

Co-authored-by: Jakob Willforss <[email protected]>

Author: ghuls

d4-hts/src/alignment/alignment_ext.rs

@@ -29,7 +29,8 @@ impl<'a> Alignment<'a> {
     }
 
     pub fn ref_end(&self) -> usize {
-        self.ref_begin() + self.ref_len()
+        // d4 uses 0-based closed interval end coordinate.
+        self.ref_begin() + self.ref_len() - 1
     }
 
     pub fn ref_id(&self) -> i32 {

d4-hts/src/lib.rs

@@ -54,9 +54,6 @@ impl<'a, R: AlignmentReader<'a>> Iterator for DepthIter<'a, R> {
             return None;
         }
 
-        let ret = (self.cur_chrom, self.cur_pos, self.heap.len() as u32);
-
-        self.cur_pos += 1;
 
         while let Some((tid, left, right)) = self.next_read {
             if tid != self.cur_chrom {
@@ -77,6 +74,10 @@ impl<'a, R: AlignmentReader<'a>> Iterator for DepthIter<'a, R> {
             self.heap.pop();
         }
 
+        let ret = (self.cur_chrom, self.cur_pos, self.heap.len() as u32);
+
+        self.cur_pos += 1;
+
         Some(ret)
     }
 }

d4tools/test/create/from-bam/output.txt

@@ -5,56 +5,56 @@
 1	10001	10002	55
 1	10002	10003	72
 1	10003	10004	93
-1	10004	10044	94
-1	10044	10046	93
-1	10046	10047	91
-1	10047	10048	90
-1	10048	10057	89
-1	10057	10059	88
-1	10059	10061	87
-1	10061	10062	86
-1	10062	10063	85
-1	10063	10064	83
-1	10064	10065	82
-1	10065	10068	78
-1	10068	10069	77
-1	10069	10071	75
-1	10071	10073	74
-1	10073	10074	71
-1	10074	10080	70
-1	10080	10081	69
-1	10081	10084	68
-1	10084	10085	67
-1	10085	10088	66
-1	10088	10089	65
-1	10089	10092	64
-1	10092	10094	62
-1	10094	10095	60
-1	10095	10102	59
-1	10102	10103	58
-1	10103	10104	54
-1	10104	10105	53
-1	10105	10106	52
-1	10106	10107	51
-1	10107	10109	49
-1	10109	10111	47
-1	10111	10113	42
-1	10113	10114	41
-1	10114	10120	36
-1	10120	10121	34
-1	10121	10126	31
-1	10126	10131	29
-1	10131	10132	28
-1	10132	10137	27
-1	10137	10138	25
-1	10138	10139	23
-1	10139	10142	21
-1	10142	10144	20
-1	10144	10146	17
-1	10146	10147	15
-1	10147	10148	14
-1	10148	10149	11
-1	10149	10150	4
-1	10150	10151	3
-1	10151	10158	1
-1	10158	20000	0
+1	10004	10043	94
+1	10043	10045	93
+1	10045	10046	91
+1	10046	10047	90
+1	10047	10056	89
+1	10056	10058	88
+1	10058	10060	87
+1	10060	10061	86
+1	10061	10062	85
+1	10062	10063	83
+1	10063	10064	82
+1	10064	10067	78
+1	10067	10068	77
+1	10068	10070	75
+1	10070	10072	74
+1	10072	10073	71
+1	10073	10079	70
+1	10079	10080	69
+1	10080	10083	68
+1	10083	10084	67
+1	10084	10087	66
+1	10087	10088	65
+1	10088	10091	64
+1	10091	10093	62
+1	10093	10094	60
+1	10094	10101	59
+1	10101	10102	58
+1	10102	10103	54
+1	10103	10104	53
+1	10104	10105	52
+1	10105	10106	51
+1	10106	10108	49
+1	10108	10110	47
+1	10110	10112	42
+1	10112	10113	41
+1	10113	10119	36
+1	10119	10120	34
+1	10120	10125	31
+1	10125	10130	29
+1	10130	10131	28
+1	10131	10136	27
+1	10136	10137	25
+1	10137	10138	23
+1	10138	10141	21
+1	10141	10143	20
+1	10143	10145	17
+1	10145	10146	15
+1	10146	10147	14
+1	10147	10148	11
+1	10148	10149	4
+1	10149	10150	3
+1	10150	10157	1
+1	10157	20000	0

d4tools/test/create/multiple-sequences-check/cmdline

@@ -0,0 +1 @@
+${D4TOOLS} create --mapping-qual 0 multi_reads_multi_chr.bam ${OUTDIR}/multi_reads_multi_chr.d4 && ${D4TOOLS} view ${OUTDIR}/multi_reads_multi_chr.d4

d4tools/test/create/multiple-sequences-check/multi_reads_multi_chr.bam

@@ -0,0 +1,14 @@
+chr1	0	4	2
+chr1	4	14	3
+chr1	14	18	2
+chr1	18	26	3
+chr1	26	30	2
+chr2	0	14	3
+chr2	14	15	2
+chr2	15	19	0
+chr2	19	22	1
+chr2	22	23	2
+chr2	23	25	3
+chr2	25	38	4
+chr2	38	39	3
+chr2	39	40	2

d4tools/test/create/multiple-sequences-check/multi_reads_multi_chr.bam.bai

@@ -0,0 +1 @@
+${D4TOOLS} create --mapping-qual 0 single_read.bam ${OUTDIR}/single_read.d4 && ${D4TOOLS} view ${OUTDIR}/single_read.d4

d4tools/test/create/multiple-sequences-check/output.txt

@@ -0,0 +1,3 @@
+chr1	0	4	0
+chr1	4	25	1
+chr1	25	30	0