示例#1
0
int bam_mpileup(int argc, char *argv[])
{
    int c;
    const char *file_list = NULL;
    char **fn = NULL;
    int nfiles = 0, use_orphan = 0, noref = 0;
    mplp_conf_t mplp;
    memset(&mplp, 0, sizeof(mplp_conf_t));
    mplp.min_baseQ = 13;
    mplp.capQ_thres = 0;
    mplp.max_depth = 250; mplp.max_indel_depth = 250;
    mplp.openQ = 40; mplp.extQ = 20; mplp.tandemQ = 100;
    mplp.min_frac = 0.002; mplp.min_support = 1;
    mplp.flag = MPLP_NO_ORPHAN | MPLP_REALN | MPLP_SMART_OVERLAPS;
    mplp.argc = argc; mplp.argv = argv;
    mplp.rflag_filter = BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP;
    mplp.output_fname = NULL;
    mplp.output_type = FT_VCF;
    mplp.record_cmd_line = 1;
    mplp.n_threads = 0;
    mplp.bsmpl = bam_smpl_init();

    static const struct option lopts[] =
    {
        {"rf", required_argument, NULL, 1},   // require flag
        {"ff", required_argument, NULL, 2},   // filter flag
        {"incl-flags", required_argument, NULL, 1},
        {"excl-flags", required_argument, NULL, 2},
        {"output", required_argument, NULL, 3},
        {"open-prob", required_argument, NULL, 4},
        {"ignore-RG", no_argument, NULL, 5},
        {"ignore-rg", no_argument, NULL, 5},
        {"gvcf", required_argument, NULL, 'g'},
        {"no-reference", no_argument, NULL, 7},
        {"no-version", no_argument, NULL, 8},
        {"threads",required_argument,NULL,9},
        {"illumina1.3+", no_argument, NULL, '6'},
        {"count-orphans", no_argument, NULL, 'A'},
        {"bam-list", required_argument, NULL, 'b'},
        {"no-BAQ", no_argument, NULL, 'B'},
        {"no-baq", no_argument, NULL, 'B'},
        {"adjust-MQ", required_argument, NULL, 'C'},
        {"adjust-mq", required_argument, NULL, 'C'},
        {"max-depth", required_argument, NULL, 'd'},
        {"redo-BAQ", no_argument, NULL, 'E'},
        {"redo-baq", no_argument, NULL, 'E'},
        {"fasta-ref", required_argument, NULL, 'f'},
        {"read-groups", required_argument, NULL, 'G'},
        {"region", required_argument, NULL, 'r'},
        {"regions", required_argument, NULL, 'r'},
        {"regions-file", required_argument, NULL, 'R'},
        {"targets", required_argument, NULL, 't'},
        {"targets-file", required_argument, NULL, 'T'},
        {"min-MQ", required_argument, NULL, 'q'},
        {"min-mq", required_argument, NULL, 'q'},
        {"min-BQ", required_argument, NULL, 'Q'},
        {"min-bq", required_argument, NULL, 'Q'},
        {"ignore-overlaps", no_argument, NULL, 'x'},
        {"output-type", required_argument, NULL, 'O'},
        {"samples", required_argument, NULL, 's'},
        {"samples-file", required_argument, NULL, 'S'},
        {"annotate", required_argument, NULL, 'a'},
        {"ext-prob", required_argument, NULL, 'e'},
        {"gap-frac", required_argument, NULL, 'F'},
        {"tandem-qual", required_argument, NULL, 'h'},
        {"skip-indels", no_argument, NULL, 'I'},
        {"max-idepth", required_argument, NULL, 'L'},
        {"min-ireads ", required_argument, NULL, 'm'},
        {"per-sample-mF", no_argument, NULL, 'p'},
        {"per-sample-mf", no_argument, NULL, 'p'},
        {"platforms", required_argument, NULL, 'P'},
        {NULL, 0, NULL, 0}
    };
    while ((c = getopt_long(argc, argv, "Ag:f:r:R:q:Q:C:Bd:L:b:P:po:e:h:Im:F:EG:6O:xa:s:S:t:T:",lopts,NULL)) >= 0) {
        switch (c) {
        case 'x': mplp.flag &= ~MPLP_SMART_OVERLAPS; break;
        case  1 :
            mplp.rflag_require = bam_str2flag(optarg);
            if ( mplp.rflag_require<0 ) { fprintf(stderr,"Could not parse --rf %s\n", optarg); return 1; }
            break;
        case  2 :
            mplp.rflag_filter = bam_str2flag(optarg);
            if ( mplp.rflag_filter<0 ) { fprintf(stderr,"Could not parse --ff %s\n", optarg); return 1; }
            break;
        case  3 : mplp.output_fname = optarg; break;
        case  4 : mplp.openQ = atoi(optarg); break;
        case  5 : bam_smpl_ignore_readgroups(mplp.bsmpl); break;
        case 'g':
            mplp.gvcf = gvcf_init(optarg);
            if ( !mplp.gvcf ) error("Could not parse: --gvcf %s\n", optarg);
            break;
        case 'f':
            mplp.fai = fai_load(optarg);
            if (mplp.fai == NULL) return 1;
            mplp.fai_fname = optarg;
            break;
        case  7 : noref = 1; break;
        case  8 : mplp.record_cmd_line = 0; break;
        case  9 : mplp.n_threads = strtol(optarg, 0, 0); break;
        case 'd': mplp.max_depth = atoi(optarg); break;
        case 'r': mplp.reg_fname = strdup(optarg); break;
        case 'R': mplp.reg_fname = strdup(optarg); mplp.reg_is_file = 1; break;
        case 't':
                  // In the original version the whole BAM was streamed which is inefficient
                  //  with few BED intervals and big BAMs. Todo: devise a heuristic to determine
                  //  best strategy, that is streaming or jumping.
                  if ( optarg[0]=='^' ) optarg++;
                  else mplp.bed_logic = 1;
                  mplp.bed = regidx_init(NULL,regidx_parse_reg,NULL,0,NULL);
                  mplp.bed_itr = regitr_init(mplp.bed);
                  if ( regidx_insert_list(mplp.bed,optarg,',') !=0 )
                  {
                      fprintf(stderr,"Could not parse the targets: %s\n", optarg);
                      exit(EXIT_FAILURE);
                  }
                  break;
        case 'T':
                  if ( optarg[0]=='^' ) optarg++;
                  else mplp.bed_logic = 1;
                  mplp.bed = regidx_init(optarg,NULL,NULL,0,NULL);
                  if (!mplp.bed) { fprintf(stderr, "bcftools mpileup: Could not read file \"%s\"", optarg); return 1; }
                  break;
        case 'P': mplp.pl_list = strdup(optarg); break;
        case 'p': mplp.flag |= MPLP_PER_SAMPLE; break;
        case 'B': mplp.flag &= ~MPLP_REALN; break;
        case 'I': mplp.flag |= MPLP_NO_INDEL; break;
        case 'E': mplp.flag |= MPLP_REDO_BAQ; break;
        case '6': mplp.flag |= MPLP_ILLUMINA13; break;
        case 's': if ( bam_smpl_add_samples(mplp.bsmpl,optarg,0)<0 ) error("Could not read samples: %s\n",optarg); break;
        case 'S': if ( bam_smpl_add_samples(mplp.bsmpl,optarg,1)<0 ) error("Could not read samples: %s\n",optarg); break;
        case 'O': 
            switch (optarg[0]) {
                case 'b': mplp.output_type = FT_BCF_GZ; break;
                case 'u': mplp.output_type = FT_BCF; break;
                case 'z': mplp.output_type = FT_VCF_GZ; break;
                case 'v': mplp.output_type = FT_VCF; break;
                default: error("[error] The option \"-O\" changed meaning when mpileup moved to bcftools. Did you mean: \"bcftools mpileup --output-type\" or \"samtools mpileup --output-BP\"?\n"); 
            }
            break;
        case 'C': mplp.capQ_thres = atoi(optarg); break;
        case 'q': mplp.min_mq = atoi(optarg); break;
        case 'Q': mplp.min_baseQ = atoi(optarg); break;
        case 'b': file_list = optarg; break;
        case 'o': {
                char *end;
                long value = strtol(optarg, &end, 10);
                // Distinguish between -o INT and -o FILE (a bit of a hack!)
                if (*end == '\0') mplp.openQ = value;
                else mplp.output_fname = optarg;
            }
            break;
        case 'e': mplp.extQ = atoi(optarg); break;
        case 'h': mplp.tandemQ = atoi(optarg); break;
        case 'A': use_orphan = 1; break;
        case 'F': mplp.min_frac = atof(optarg); break;
        case 'm': mplp.min_support = atoi(optarg); break;
        case 'L': mplp.max_indel_depth = atoi(optarg); break;
        case 'G': bam_smpl_add_readgroups(mplp.bsmpl, optarg, 1); break;
        case 'a':
            if (optarg[0]=='?') {
                list_annotations(stderr);
                return 1;
            }
            mplp.fmt_flag |= parse_format_flag(optarg);
        break;
        default:
            fprintf(stderr,"Invalid option: '%c'\n", c);
            return 1;
        }
    }

    if ( mplp.gvcf && !(mplp.fmt_flag&B2B_FMT_DP) )
    {
        fprintf(stderr,"[warning] The -t DP option is required with --gvcf, switching on.\n");
        mplp.fmt_flag |= B2B_FMT_DP;
    }
    if ( mplp.flag&(MPLP_BCF|MPLP_VCF|MPLP_NO_COMP) )
    {
        if ( mplp.flag&MPLP_VCF )
        {
            if ( mplp.flag&MPLP_NO_COMP ) mplp.output_type = FT_VCF;
            else mplp.output_type = FT_VCF_GZ;
        }
        else if ( mplp.flag&MPLP_BCF )
        {
            if ( mplp.flag&MPLP_NO_COMP ) mplp.output_type = FT_BCF;
            else mplp.output_type = FT_BCF_GZ;
        }
    }
    if ( !(mplp.flag&MPLP_REALN) && mplp.flag&MPLP_REDO_BAQ )
    {
        fprintf(stderr,"Error: The -B option cannot be combined with -E\n");
        return 1;
    }
    if (use_orphan) mplp.flag &= ~MPLP_NO_ORPHAN;
    if (argc == 1)
    {
        print_usage(stderr, &mplp);
        return 1;
    }
    if (!mplp.fai && !noref) {
        fprintf(stderr,"Error: mpileup requires the --fasta-ref option by default; use --no-reference to run without a fasta reference\n");
        return 1;
    }
    int ret,i;
    if (file_list) 
    {
        if ( read_file_list(file_list,&nfiles,&fn) ) return 1;
        mplp.files  = fn;
        mplp.nfiles = nfiles;
    }
    else
    {
        mplp.nfiles = argc - optind;
        mplp.files  = (char**) malloc(mplp.nfiles*sizeof(char*));
        for (i=0; i<mplp.nfiles; i++) mplp.files[i] = strdup(argv[optind+i]);
    }
    ret = mpileup(&mplp);

    for (i=0; i<mplp.nfiles; i++) free(mplp.files[i]);
    free(mplp.files);
    free(mplp.reg_fname); free(mplp.pl_list);
    if (mplp.fai) fai_destroy(mplp.fai);
    if (mplp.bed) regidx_destroy(mplp.bed);
    if (mplp.bed_itr) regitr_destroy(mplp.bed_itr);
    if (mplp.reg) regidx_destroy(mplp.reg);
    bam_smpl_destroy(mplp.bsmpl);
    return ret;
}
示例#2
0
int main(int argc, char *argv[])
{
   int first,i;
   char* f;
   group_t* group;
   int sh_override = -1;
   int list_packages = 0;

   for (i=1; i<argc && *argv[i] == '-'; i++)
   {
      for (f=argv[i]+1; *f; f++)
         switch (*f)
         {
            case 'v':
               debugging = 1;
               silent = 0;
               break;
            case 's':
               debugging = 0;
               silent = 1;
               break;
            case 'c':
               sh_override = 1;
               break;
            case 'b':
               sh_override = 0;
               break;
            case 'f':
               main_package_filename = argv[++i];
               break;
            case 'l':
               list_packages = 1;
               break;
            default:
               fprintf(stderr, "usepackage: unrecognised flag '%c'\n", *f);
               exit(1);
         }
   }

   if (!list_packages && (i >= argc))
   {
      fprintf(stderr, "\n%s %s, %s\n", PACKAGE_NAME, PACKAGE_VERSION, COPYRIGHT);
      fprintf(stderr, "Bug reports and comments to: %s\n", PACKAGE_BUGREPORT);
      fprintf(stderr, "Discussion/announcement list: %s\n\n", MAILING_LIST);
      fprintf(stderr, "usage: use [-vs] [-f <file>] <package> [<package>...]\n");
      fprintf(stderr, "       use -l\n\n");
      fprintf(stderr, "       -v : verbose\n");
      fprintf(stderr, "       -s : silence match warnings\n");
      fprintf(stderr, "       -f : use <file> as main packages file\n");
      fprintf(stderr, "       -l : list available packages\n\n");
      exit(1);
   }

   DEBUG(stderr, "%s\n", PACKAGE_NAME);
   DEBUG(stderr, "Version: %s\n", PACKAGE_VERSION);
   DEBUG(stderr, "%s\n", COPYRIGHT);

   uname(&the_host_info);
   DEBUG(stderr, "host: %s\n", the_host_info.nodename);
   DEBUG(stderr, "operating system: %s %s\n", the_host_info.sysname,
         the_host_info.release);
   DEBUG(stderr, "architecture: %s\n", the_host_info.machine);

   the_shell = get_user_shell();
   DEBUG(stderr, "shell: %s\n", the_shell);

   if (sh_override != -1)
      csh_user = sh_override;
   else
      csh_user = ((!strcmp(the_shell, "csh")) || (!strcmp(the_shell, "tcsh")));

   the_environment = new_list();
   the_scripts = new_list();
   
   if (get_packages(&the_packages, &the_groups, &the_annotations))
   {
      fprintf(stderr, "usepackage: couldn't load package information.\n");
      exit(2);
   }

   if (list_packages)
   {
      fprintf(stderr, "\nusepackage %s, %s\n\n", VERSION, COPYRIGHT);
      fprintf(stderr, "Available packages are:\n\n");
      list_annotations();
      fprintf(stderr, "\nAvailable groups are:\n\n");
      list_groups();
      fprintf(stderr, "\n");
      exit(0);
   }

   for (first = i, i = argc - 1 ; i >= first ; i--)
   {
      if (group = get_group(argv[i]))
         use_group(group);
      else
         use_package(argv[i]);
   }

   print_env();

   print_scripts();

   return(0);
}