src/hg/lib/bamFile.c 1.2

Index: src/hg/lib/bamFile.c
===================================================================
RCS file: /projects/compbio/cvsroot/kent/src/hg/lib/bamFile.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -b -B -U 1000000 -r1.1 -r1.2
--- src/hg/lib/bamFile.c	27 Jul 2009 21:52:08 -0000	1.1
+++ src/hg/lib/bamFile.c	3 Aug 2009 22:00:24 -0000	1.2
@@ -1,52 +1,117 @@
 /* bam -- interface to binary alignment format files using Heng Li's samtools lib. */
 
 #ifdef USE_BAM
 
 #include "common.h"
 #include "hdb.h"
 #include "bamFile.h"
 
 static char const rcsid[] = "$Id$";
 
 static char *bbiNameFromTable(struct sqlConnection *conn, char *table)
 /* Return file name from little table. */
 /* This should be libified somewhere sensible -- copied from hgTracks/bedTrack.c. */
 {
 char query[256];
 safef(query, sizeof(query), "select fileName from %s", table);
 char *fileName = sqlQuickString(conn, query);
 if (fileName == NULL)
     errAbort("Missing fileName in %s table", table);
 return fileName;
 }
 
 void bamFetch(char *db, char *table, char *position, bam_fetch_f callbackFunc, void *callbackData)
 /* Open the .bam file given in db.table, fetch items in the seq:start-end position range,
  * and call callbackFunc on each bam item retrieved from the file plus callbackData. 
  * Note: if sequences in .bam file don't begin with "chr" but cart position does, pass in 
  * cart position + strlen("chr") to match the .bam file sequence names. */
 {
 // TODO: if bamFile is URL, convert URL to path a la UDC, but under udcFuse mountpoint.
 // new hg.conf setting for udcFuse mountpoint/support.
 struct sqlConnection *conn = hAllocConn(db);
 char *bamFileName = bbiNameFromTable(conn, table);
 hFreeConn(&conn);
 
 samfile_t *fh = samopen(bamFileName, "rb", NULL);
 if (fh == NULL)
     errAbort("samopen(%s, \"rb\") returned NULL", bamFileName);
 
 int chromId, start, end;
 int ret = bam_parse_region(fh->header, position, &chromId, &start, &end);
 if (ret != 0)
     errAbort("bam_parse_region(%s) failed (%d)", position, ret);
 //?? Could this happen if there is no data on some _random?  can avoid with tdb chromosomes...
 
 bam_index_t *idx = bam_index_load(bamFileName);
 ret = bam_fetch(fh->x.bam, idx, chromId, start, end, callbackData, callbackFunc);
 if (ret != 0)
     errAbort("bam_fetch(%s, %s (chromId=%d) failed (%d)", bamFileName, position, chromId, ret);
 samclose(fh);
 }
 
+char *bamGetQuerySequence(const bam1_t *bam)
+/* Return the nucleotide sequence encoded in bam. */
+{
+const bam1_core_t *core = &bam->core;
+char *qSeq = needMem(core->l_qseq + 1);
+uint8_t *s = bam1_seq(bam);
+int i;
+for (i = 0; i < core->l_qseq; i++)
+    qSeq[i] = bam_nt16_rev_table[bam1_seqi(s, i)];
+if ((core->flag & BAM_FREVERSE) == 1)
+    reverseComplement(qSeq, core->l_qseq);
+return qSeq;
+}
+
+char *bamGetCigar(const bam1_t *bam)
+/* Return a BAM-enhanced CIGAR string, decoded from the packed encoding in bam. */
+{
+unsigned int *cigarPacked = bam1_cigar(bam);
+const bam1_core_t *core = &bam->core;
+struct dyString *dyCigar = dyStringNew(min(8, core->n_cigar*4));
+int i;
+for (i = 0;  i < core->n_cigar;  i++)
+    {
+    char op;
+    int n = bamUnpackCigarElement(cigarPacked[i], &op);
+    dyStringPrintf(dyCigar, "%d", n);
+    dyStringAppendC(dyCigar, op);
+    }
+return dyStringCannibalize(&dyCigar);
+}
+
+int bamGetTargetLength(const bam1_t *bam)
+/* Tally up the alignment's length on the reference sequence from
+ * bam's packed-int CIGAR representation. */
+{
+unsigned int *cigarPacked = bam1_cigar(bam);
+const bam1_core_t *core = &bam->core;
+int tLength=0;
+int i;
+for (i = 0;  i < core->n_cigar;  i++)
+    {
+    char op;
+    int n = bamUnpackCigarElement(cigarPacked[i], &op);
+    switch (op)
+	{
+	case 'M': // match or mismatch (gapless aligned block)
+	    tLength += n;
+	    break;
+	case 'I': // inserted in query
+	case 'S': // skipped query bases at beginning or end ("soft clipping")
+	    break;
+	case 'D': // deleted from query
+	case 'N': // long deletion from query (intron as opposed to small del)
+	    tLength += n;
+	    break;
+	case 'H': // skipped query bases not stored in record's query sequence ("hard clipping")
+	case 'P': // P="silent deletion from padded reference sequence" -- ignore these.
+	    break;
+	default:
+	    errAbort("bamGetTargetLength: unrecognized CIGAR op %c -- update me", op);
+	}
+    }
+return tLength;
+}
+
 #endif//def USE_BAM