X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bcftools%2Fcall1.c;h=dda4164d111a55747d85f74fd814c90a4bab11d4;hb=3f820f8c565a8162d835a321a5c0ecdd9567d820;hp=415426abcd1511f1e2cadd987501efcbf21bcb74;hpb=74e6aebb5c07d14cdccc5f8abb40fd4a00b2be86;p=samtools.git

diff --git a/bcftools/call1.c b/bcftools/call1.c
index 415426a..dda4164 100644
--- a/bcftools/call1.c
+++ b/bcftools/call1.c
@@ -48,11 +48,6 @@ void *bed_read(const char *fn);
 void bed_destroy(void *_h);
 int bed_overlap(const void *_h, const char *chr, int beg, int end);
 
-typedef struct {
-	double p[4];
-	int mq, depth, is_tested, d[4];
-} anno16_t;
-
 static double ttest(int n1, int n2, int a[4])
 {
 	extern double kf_betai(double a, double b, double x);
@@ -83,7 +78,7 @@ static int test16_core(int anno[16], anno16_t *a)
 	return 0;
 }
 
-static int test16(bcf1_t *b, anno16_t *a)
+int test16(bcf1_t *b, anno16_t *a)
 {
 	char *p;
 	int i, anno[16];
@@ -100,17 +95,6 @@ static int test16(bcf1_t *b, anno16_t *a)
 	return test16_core(anno, a);
 }
 
-static void rm_info(bcf1_t *b, const char *key)
-{
-	char *p, *q;
-	if ((p = strstr(b->info, key)) == 0) return;
-	for (q = p; *q && *q != ';'; ++q);
-	if (p > b->info && *(p-1) == ';') --p;
-	memmove(p, q, b->l_str - (q - b->str));
-	b->l_str -= q - p;
-	bcf_sync(b);
-}
-
 static int update_bcf1(bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr, double pref, int flag, double em[10], int cons_llr, int64_t cons_gt)
 {
 	kstring_t s;
@@ -119,7 +103,7 @@ static int update_bcf1(bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr,
 	anno16_t a;
 
 	has_I16 = test16(b, &a) >= 0? 1 : 0;
-	rm_info(b, "I16="); // FIXME: probably this function has a bug. If I move it below, I16 will not be removed!
+	//rm_info(b, "I16="); // FIXME: probably this function has a bug. If I move it below, I16 will not be removed!
 
 	memset(&s, 0, sizeof(kstring_t));
 	kputc('\0', &s); kputs(b->ref, &s); kputc('\0', &s);
@@ -170,6 +154,8 @@ static int update_bcf1(bcf1_t *b, const bcf_p1aux_t *pa, const bcf_p1rst_t *pr,
 	}
 	if (has_I16 && a.is_tested) ksprintf(&s, ";PV4=%.2g,%.2g,%.2g,%.2g", a.p[0], a.p[1], a.p[2], a.p[3]);
 	kputc('\0', &s);
+    rm_info(&s, "QS=");
+    rm_info(&s, "I16=");
 	kputs(b->fmt, &s); kputc('\0', &s);
 	free(b->str);
 	b->m_str = s.m; b->l_str = s.l; b->str = s.s;
@@ -312,6 +298,9 @@ int bcfview(int argc, char *argv[])
 	extern int bcf_trio_call(uint32_t *prep, const bcf1_t *b, int *llr, int64_t *gt);
 	extern int bcf_pair_call(const bcf1_t *b);
 	extern int bcf_min_diff(const bcf1_t *b);
+	extern int bcf_p1_get_M(bcf_p1aux_t *b);
+
+	extern gzFile bcf_p1_fp_lk;
 
 	bcf_t *bp, *bout = 0;
 	bcf1_t *b, *blast;
@@ -327,10 +316,10 @@ int bcfview(int argc, char *argv[])
 	memset(&vc, 0, sizeof(viewconf_t));
 	vc.prior_type = vc.n1 = -1; vc.theta = 1e-3; vc.pref = 0.5; vc.indel_frac = -1.; vc.n_perm = 0; vc.min_perm_p = 0.01; vc.min_smpl_frac = 0; vc.min_lrt = 1; vc.min_ma_lrt = -1;
 	memset(qcnt, 0, 8 * 256);
-	while ((c = getopt(argc, argv, "FN1:l:cC:eHAGvbSuP:t:p:QgLi:IMs:D:U:X:d:T:Ywm:")) >= 0) {
+	while ((c = getopt(argc, argv, "FN1:l:cC:eHAGvbSuP:t:p:QgLi:IMs:D:U:X:d:T:Ywm:K:")) >= 0) {
 		switch (c) {
 		case '1': vc.n1 = atoi(optarg); break;
-		case 'l': vc.bed = bed_read(optarg); break;
+		case 'l': vc.bed = bed_read(optarg); if (!vc.bed) { fprintf(stderr,"Could not read \"%s\"\n", optarg); return 1; } break;
 		case 'D': vc.fn_dict = strdup(optarg); break;
 		case 'F': vc.flag |= VC_FIX_PL; break;
 		case 'N': vc.flag |= VC_ACGT_ONLY; break;
@@ -357,6 +346,7 @@ int bcfview(int argc, char *argv[])
 		case 'C': vc.min_lrt = atof(optarg); break;
 		case 'X': vc.min_perm_p = atof(optarg); break;
 		case 'd': vc.min_smpl_frac = atof(optarg); break;
+		case 'K': bcf_p1_fp_lk = gzopen(optarg, "w"); break;
 		case 's': vc.subsam = read_samples(optarg, &vc.n_sub);
 			vc.ploidy = calloc(vc.n_sub + 1, 1);
 			for (tid = 0; tid < vc.n_sub; ++tid) vc.ploidy[tid] = vc.subsam[tid][strlen(vc.subsam[tid]) + 1];
@@ -446,7 +436,7 @@ int bcfview(int argc, char *argv[])
 			vc.sublist = calloc(vc.n_sub, sizeof(int));
 			hout = bcf_hdr_subsam(hin, vc.n_sub, vc.subsam, vc.sublist);
 		}
-		if (vc.flag & VC_CALL) write_header(hout);
+		write_header(hout); // always print the header
 		vcf_hdr_write(bout, hout);
 	}
 	if (vc.flag & VC_CALL) {
@@ -480,6 +470,10 @@ int bcfview(int argc, char *argv[])
 			}
 		}
 	}
+	if (bcf_p1_fp_lk && p1) {
+		int32_t M = bcf_p1_get_M(p1);
+		gzwrite(bcf_p1_fp_lk, &M, 4);
+	}
 	while (vcf_read(bp, hin, b) > 0) {
 		int is_indel, cons_llr = -1;
 		int64_t cons_gt = -1;
@@ -528,15 +522,19 @@ int bcfview(int argc, char *argv[])
 			int i;
 			for (i = 0; i < 9; ++i) em[i] = -1.;
 		}
-        bcf_p1_set_ploidy(b, p1); // could be improved: do this per site to allow pseudo-autosomal regions
-        if ( !(vc.flag&VC_KEEPALT) && vc.flag&VC_CALL && vc.min_ma_lrt>=0 )
+        if ( !(vc.flag&VC_KEEPALT) && (vc.flag&VC_CALL) && vc.min_ma_lrt>=0 )
         {
+            bcf_p1_set_ploidy(b, p1); // could be improved: do this per site to allow pseudo-autosomal regions
             int gts = call_multiallelic_gt(b,p1,vc.min_ma_lrt);
             if ( gts<=1 && vc.flag & VC_VARONLY ) continue;
         }
 		else if (vc.flag & VC_CALL) { // call variants
 			bcf_p1rst_t pr;
-			int calret = bcf_p1_cal(b, (em[7] >= 0 && em[7] < vc.min_lrt), p1, &pr);
+			int calret;
+			gzwrite(bcf_p1_fp_lk, &b->tid, 4);
+			gzwrite(bcf_p1_fp_lk, &b->pos, 4);
+			gzwrite(bcf_p1_fp_lk, &em[0], sizeof(double));
+			calret = bcf_p1_cal(b, (em[7] >= 0 && em[7] < vc.min_lrt), p1, &pr);
 			if (n_processed % 100000 == 0) {
 				fprintf(stderr, "[%s] %ld sites processed.\n", __func__, (long)n_processed);
 				bcf_p1_dump_afs(p1);
@@ -581,6 +579,8 @@ int bcfview(int argc, char *argv[])
 		} else bcf_fix_gt(b);
 		vcf_write(bout, hout, b);
 	}
+
+	if (bcf_p1_fp_lk) gzclose(bcf_p1_fp_lk);
 	if (vc.prior_file) free(vc.prior_file);
 	if (vc.flag & VC_CALL) bcf_p1_dump_afs(p1);
 	if (hin != hout) bcf_hdr_destroy(hout);