X-Git-Url: https://git.donarmstrong.com/?p=samtools.git;a=blobdiff_plain;f=phase.c;h=ef4eff952bed637e4e9f89ad1b8c10619869c59f;hp=5fc4ce77d94ed3046082abf7a9525caf3362f7c3;hb=307c147168f7154e3755712797078c513e0b242a;hpb=fe25084a4eac12a94042805e6529ad19a2b70c28

diff --git a/phase.c b/phase.c
index 5fc4ce7..ef4eff9 100644
--- a/phase.c
+++ b/phase.c
@@ -17,6 +17,7 @@ KSTREAM_INIT(gzFile, gzread, 16384)
 
 #define FLAG_FIX_CHIMERA 0x1
 #define FLAG_LIST_EXCL   0x4
+#define FLAG_DROP_AMBI   0x8
 
 typedef struct {
 	// configurations, initialized in the main function
@@ -34,7 +35,7 @@ typedef struct {
 typedef struct {
 	int8_t seq[MAX_VARS]; // TODO: change to dynamic memory allocation!
 	int vpos, beg, end;
-	uint32_t vlen:16, single:1, flip:1, phase:1, phased:1;
+	uint32_t vlen:16, single:1, flip:1, phase:1, phased:1, ambig:1;
 	uint32_t in:16, out:16; // in-phase and out-phase
 } frag_t, *frag_p;
 
@@ -176,7 +177,8 @@ static uint64_t *fragphase(int vpos, const int8_t *path, nseq_t *hash, int flip)
 			}
 			f->phase = c[0] > c[1]? 0 : 1;
 			f->in = c[f->phase]; f->out = c[1 - f->phase];
-			if (f->in && f->out && f->in <= f->out + 1) f->phased = 0;
+			f->phased = f->in == f->out? 0 : 1;
+			f->ambig = (f->in && f->out && f->out < 3 && f->in <= f->out + 1)? 1 : 0;
 			// fix chimera
 			f->flip = 0;
 			if (flip && c[0] >= 3 && c[1] >= 3) {
@@ -307,7 +309,8 @@ static int clean_seqs(int vpos, nseq_t *hash)
 
 static void dump_aln(phaseg_t *g, int min_pos, const nseq_t *hash)
 {
-	int i, is_flip;
+	int i, is_flip, drop_ambi;
+	drop_ambi = g->flag & FLAG_DROP_AMBI;
 	is_flip = (drand48() < 0.5);
 	for (i = 0; i < g->n; ++i) {
 		int end, which;
@@ -321,8 +324,9 @@ static void dump_aln(phaseg_t *g, int min_pos, const nseq_t *hash)
 		if (k == kh_end(hash)) which = 3;
 		else {
 			frag_t *f = &kh_val(hash, k);
-			if (f->phased && f->flip) which = 2;
-			else if (f->phased == 0) which = 2;
+			if (f->ambig) which = drop_ambi? 2 : 3;
+			else if (f->phased && f->flip) which = 2;
+			else if (f->phased == 0) which = 3;
 			else { // phased and not flipped
 				char c = 'Y';
 				which = f->phase;
@@ -519,7 +523,7 @@ int main_phase(int argc, char *argv[])
 	memset(&g, 0, sizeof(phaseg_t));
 	g.flag = FLAG_FIX_CHIMERA;
 	g.min_varLOD = 37; g.k = 13; g.min_baseQ = 13; g.max_depth = 256;
-	while ((c = getopt(argc, argv, "Q:eFq:k:b:l:D:")) >= 0) {
+	while ((c = getopt(argc, argv, "Q:eFq:k:b:l:D:A:")) >= 0) {
 		switch (c) {
 			case 'D': g.max_depth = atoi(optarg); break;
 			case 'q': g.min_varLOD = atoi(optarg); break;
@@ -527,6 +531,7 @@ int main_phase(int argc, char *argv[])
 			case 'k': g.k = atoi(optarg); break;
 			case 'F': g.flag &= ~FLAG_FIX_CHIMERA; break;
 			case 'e': g.flag |= FLAG_LIST_EXCL; break;
+			case 'A': g.flag |= FLAG_DROP_AMBI; break;
 			case 'b': g.pre = strdup(optarg); break;
 			case 'l': fn_list = strdup(optarg); break;
 		}
@@ -541,6 +546,7 @@ int main_phase(int argc, char *argv[])
 		fprintf(stderr, "         -D INT    max read depth [%d]\n", g.max_depth);
 //		fprintf(stderr, "         -l FILE   list of sites to phase [null]\n");
 		fprintf(stderr, "         -F        do not attempt to fix chimeras\n");
+		fprintf(stderr, "         -A        drop reads with ambiguous phase\n");
 //		fprintf(stderr, "         -e        do not discover SNPs (effective with -l)\n");
 		fprintf(stderr, "\n");
 		return 1;
@@ -652,7 +658,7 @@ int main_phase(int argc, char *argv[])
 				memset(f->seq, 0, MAX_VARS);
 				f->beg = p->b->core.pos;
 				f->end = bam_calend(&p->b->core, bam1_cigar(p->b));
-				f->vpos = vpos, f->vlen = 1, f->seq[0] = c, f->single = f->phased = f->flip = 0;
+				f->vpos = vpos, f->vlen = 1, f->seq[0] = c, f->single = f->phased = f->flip = f->ambig = 0;
 			}
 		}
 		if (dophase) {