X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sam_view.c;h=c7a47747b4aba916911534591411ae0721b962a2;hb=0b3418cf166ce4a58cedf0d9a2df5ec3dd4cc5fa;hp=f198591e9aabbe44f67ee35c1497d2e73d0a3d7d;hpb=bbf6f360e153cf98d091afdaa31717448cde21e9;p=samtools.git diff --git a/sam_view.c b/sam_view.c index f198591..c7a4774 100644 --- a/sam_view.c +++ b/sam_view.c @@ -21,8 +21,9 @@ typedef khash_t(rg) *rghash_t; // FIXME: we'd better use no global variables... static rghash_t g_rghash = 0; -static int g_min_mapQ = 0, g_flag_on = 0, g_flag_off = 0, g_qual_scale = 0; -static float g_subsam = -1; +static int g_min_mapQ = 0, g_flag_on = 0, g_flag_off = 0, g_qual_scale = 0, g_min_qlen = 0; +static uint32_t g_subsam_seed = 0; +static double g_subsam_frac = -1.; static char *g_library, *g_rg; static void *g_bed; @@ -40,14 +41,21 @@ static int process_aln(const bam_header_t *h, bam1_t *b) qual[i] = c < 93? c : 93; } } + if (g_min_qlen > 0) { + int k, qlen = 0; + uint32_t *cigar = bam1_cigar(b); + for (k = 0; k < b->core.n_cigar; ++k) + if ((bam_cigar_type(bam_cigar_op(cigar[k]))&1) || bam_cigar_op(cigar[k]) == BAM_CHARD_CLIP) + qlen += bam_cigar_oplen(cigar[k]); + if (qlen < g_min_qlen) return 1; + } if (b->core.qual < g_min_mapQ || ((b->core.flag & g_flag_on) != g_flag_on) || (b->core.flag & g_flag_off)) return 1; if (g_bed && b->core.tid >= 0 && !bed_overlap(g_bed, h->target_name[b->core.tid], b->core.pos, bam_calend(&b->core, bam1_cigar(b)))) return 1; - if (g_subsam > 0.) { - int x = (int)(g_subsam + .499); - uint32_t k = __ac_X31_hash_string(bam1_qname(b)) + x; - if (k%1024 / 1024.0 >= g_subsam - x) return 1; + if (g_subsam_frac > 0.) { + uint32_t k = __ac_X31_hash_string(bam1_qname(b)) + g_subsam_seed; + if ((double)(k&0xffffff) / 0x1000000 >= g_subsam_frac) return 1; } if (g_rg || g_rghash) { uint8_t *s = bam_aux_get(b, "RG"); @@ -122,13 +130,20 @@ int main_samview(int argc, char *argv[]) int of_type = BAM_OFDEC, is_long_help = 0, n_threads = 0; int count = 0; samfile_t *in = 0, *out = 0; - char in_mode[5], out_mode[5], *fn_out = 0, *fn_list = 0, *fn_ref = 0, *fn_rg = 0; + char in_mode[5], out_mode[5], *fn_out = 0, *fn_list = 0, *fn_ref = 0, *fn_rg = 0, *q; /* parse command-line options */ strcpy(in_mode, "r"); strcpy(out_mode, "w"); - while ((c = getopt(argc, argv, "SbBct:h1Ho:q:f:F:ul:r:xX?T:R:L:s:Q:@:")) >= 0) { + while ((c = getopt(argc, argv, "SbBct:h1Ho:q:f:F:ul:r:xX?T:R:L:s:Q:@:m:")) >= 0) { switch (c) { - case 's': g_subsam = atof(optarg); break; + case 's': + if ((g_subsam_seed = strtol(optarg, &q, 10)) != 0) { + srand(g_subsam_seed); + g_subsam_seed = rand(); + } + g_subsam_frac = strtod(q, &q); + break; + case 'm': g_min_qlen = atoi(optarg); break; case 'c': is_count = 1; break; case 'S': is_bamin = 0; break; case 'b': is_bamout = 1; break; @@ -290,6 +305,7 @@ static int usage(int is_long_help) fprintf(stderr, " -X output FLAG in string (samtools-C specific)\n"); fprintf(stderr, " -c print only the count of matching records\n"); fprintf(stderr, " -B collapse the backward CIGAR operation\n"); + fprintf(stderr, " -@ INT number of BAM compression threads [0]\n"); fprintf(stderr, " -L FILE output alignments overlapping the input BED FILE [null]\n"); fprintf(stderr, " -t FILE list of reference names and lengths (force -S) [null]\n"); fprintf(stderr, " -T FILE reference sequence file (force -S) [null]\n"); @@ -360,12 +376,14 @@ int main_bam2fq(int argc, char *argv[]) bam_header_t *h; bam1_t *b; int8_t *buf; - int max_buf; + int max_buf, c, no12 = 0; + while ((c = getopt(argc, argv, "n")) > 0) + if (c == 'n') no12 = 1; if (argc == 1) { fprintf(stderr, "Usage: samtools bam2fq \n"); return 1; } - fp = strcmp(argv[1], "-")? bam_open(argv[1], "r") : bam_dopen(fileno(stdin), "r"); + fp = strcmp(argv[optind], "-")? bam_open(argv[optind], "r") : bam_dopen(fileno(stdin), "r"); if (fp == 0) return 1; h = bam_header_read(fp); b = bam_init1(); @@ -375,9 +393,12 @@ int main_bam2fq(int argc, char *argv[]) int i, qlen = b->core.l_qseq; uint8_t *seq; putchar('@'); fputs(bam1_qname(b), stdout); - if ((b->core.flag & 0x40) && !(b->core.flag & 0x80)) puts("/1"); - else if ((b->core.flag & 0x80) && !(b->core.flag & 0x40)) puts("/2"); - else putchar('\n'); + if (no12) putchar('\n'); + else { + if ((b->core.flag & 0x40) && !(b->core.flag & 0x80)) puts("/1"); + else if ((b->core.flag & 0x80) && !(b->core.flag & 0x40)) puts("/2"); + else putchar('\n'); + } if (max_buf < qlen + 1) { max_buf = qlen + 1; kroundup32(max_buf);