From: Heng Li <lh3@me.com>
Date: Tue, 20 Mar 2012 15:49:52 +0000 (-0400)
Subject: Merge branch 'master' into mt
X-Git-Url: https://git.donarmstrong.com/?a=commitdiff_plain;h=185e80bd42cead0e5af6ae0c54ebb6d0b66ca47c;hp=-c;p=samtools.git

Merge branch 'master' into mt
---

185e80bd42cead0e5af6ae0c54ebb6d0b66ca47c
diff --combined sam_view.c
index b6896c9,ce1e109..d1178ff
--- a/sam_view.c
+++ b/sam_view.c
@@@ -22,7 -22,8 +22,8 @@@ 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 uint32_t g_subsam_seed = 0;
+ static double g_subsam_frac = -1.;
  static char *g_library, *g_rg;
  static void *g_bed;
  
@@@ -44,10 -45,9 +45,9 @@@ static int process_aln(const bam_header
  		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");
@@@ -119,16 -119,22 +119,22 @@@ static int usage(int is_long_help)
  int main_samview(int argc, char *argv[])
  {
  	int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0, compress_level = -1, is_bamout = 0, is_count = 0;
 -	int of_type = BAM_OFDEC, is_long_help = 0;
 +	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:@:")) >= 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 'c': is_count = 1; break;
  		case 'S': is_bamin = 0; break;
  		case 'b': is_bamout = 1; break;
@@@ -151,7 -157,6 +157,7 @@@
  		case 'T': fn_ref = strdup(optarg); is_bamin = 0; break;
  		case 'B': bam_no_B = 1; break;
  		case 'Q': g_qual_scale = atoi(optarg); break;
 +		case '@': n_threads = strtol(optarg, 0, 0); break;
  		default: return usage(is_long_help);
  		}
  	}
@@@ -209,7 -214,6 +215,7 @@@
  		ret = 1;
  		goto view_end;
  	}
 +	if (n_threads > 1) samthreads(out, n_threads, 256); 
  	if (is_header_only) goto view_end; // no need to print alignments
  
  	if (argc == optind + 1) { // convert/print the entire file
@@@ -290,7 -294,6 +296,7 @@@ 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");