Пример #1
0
// Expands a output filename format string
static char* expand_format_string(const char* format_string, const char* basename, const char* rg_id, const int rg_idx)
{
    kstring_t str = { 0, 0, NULL };
    const char* pointer = format_string;
    const char* next;
    while ((next = strchr(pointer, '%')) != NULL) {
        kputsn(pointer, next-pointer, &str);
        ++next;
        switch (*next) {
        case '%':
            kputc('%', &str);
            break;
        case '*':
            kputs(basename, &str);
            break;
        case '#':
            kputl(rg_idx, &str);
            break;
        case '!':
            kputs(rg_id, &str);
            break;
        case '\0':
            // Error is: fprintf(pysamerr, "bad format string, trailing %%\n");
            free(str.s);
            return NULL;
        default:
            // Error is: fprintf(pysamerr, "bad format string, unknown format specifier\n");
            free(str.s);
            return NULL;
        }
        pointer = next + 1;
    }
    kputs(pointer, &str);
    return ks_release(&str);
}
Пример #2
0
// Expands a output filename format string
static char* expand_format_string(const char* format_string, const char* basename, const char* rg_id, const int rg_idx, const htsFormat *format)
{
    kstring_t str = { 0, 0, NULL };
    const char* pointer = format_string;
    const char* next;
    while ((next = strchr(pointer, '%')) != NULL) {
        kputsn(pointer, next-pointer, &str);
        ++next;
        switch (*next) {
            case '%':
                kputc('%', &str);
                break;
            case '*':
                kputs(basename, &str);
                break;
            case '#':
                kputl(rg_idx, &str);
                break;
            case '!':
                kputs(rg_id, &str);
                break;
            case '.':
                // Only really need to cope with sam, bam, cram
                if (format->format != unknown_format)
                    kputs(hts_format_file_extension(format), &str);
                else
                    kputs("bam", &str);
                break;
            case '\0':
                // Error is: fprintf(pysam_stderr, "bad format string, trailing %%\n");
                free(str.s);
                return NULL;
            default:
                // Error is: fprintf(pysam_stderr, "bad format string, unknown format specifier\n");
                free(str.s);
                return NULL;
        }
        pointer = next + 1;
    }
    kputs(pointer, &str);
    return ks_release(&str);
}
Пример #3
0
int main_bedcov(int argc, char *argv[])
{
    gzFile fp;
    kstring_t str;
    kstream_t *ks;
    hts_idx_t **idx;
    aux_t **aux;
    int *n_plp, dret, i, n, c, min_mapQ = 0;
    int64_t *cnt;
    const bam_pileup1_t **plp;
    int usage = 0;

    sam_global_args ga = SAM_GLOBAL_ARGS_INIT;
    static const struct option lopts[] = {
        SAM_OPT_GLOBAL_OPTIONS('-', 0, '-', '-', 0),
        { NULL, 0, NULL, 0 }
    };

    while ((c = getopt_long(argc, argv, "Q:", lopts, NULL)) >= 0) {
        switch (c) {
        case 'Q': min_mapQ = atoi(optarg); break;
        default:  if (parse_sam_global_opt(c, optarg, lopts, &ga) == 0) break;
                  /* else fall-through */
        case '?': usage = 1; break;
        }
        if (usage) break;
    }
    if (usage || optind + 2 > argc) {
        fprintf(pysam_stderr, "Usage: samtools bedcov [options] <in.bed> <in1.bam> [...]\n\n");
        fprintf(pysam_stderr, "  -Q INT       Only count bases of at least INT quality [0]\n");
        sam_global_opt_help(pysam_stderr, "-.--.");
        return 1;
    }
    memset(&str, 0, sizeof(kstring_t));
    n = argc - optind - 1;
    aux = calloc(n, sizeof(aux_t*));
    idx = calloc(n, sizeof(hts_idx_t*));
    for (i = 0; i < n; ++i) {
        aux[i] = calloc(1, sizeof(aux_t));
        aux[i]->min_mapQ = min_mapQ;
        aux[i]->fp = sam_open_format(argv[i+optind+1], "r", &ga.in);
        if (aux[i]->fp)
            idx[i] = sam_index_load(aux[i]->fp, argv[i+optind+1]);
        if (aux[i]->fp == 0 || idx[i] == 0) {
            fprintf(pysam_stderr, "ERROR: fail to open index BAM file '%s'\n", argv[i+optind+1]);
            return 2;
        }
        // TODO bgzf_set_cache_size(aux[i]->fp, 20);
        aux[i]->header = sam_hdr_read(aux[i]->fp);
        if (aux[i]->header == NULL) {
            fprintf(pysam_stderr, "ERROR: failed to read header for '%s'\n",
                    argv[i+optind+1]);
            return 2;
        }
    }
    cnt = calloc(n, 8);

    fp = gzopen(argv[optind], "rb");
    ks = ks_init(fp);
    n_plp = calloc(n, sizeof(int));
    plp = calloc(n, sizeof(bam_pileup1_t*));
    while (ks_getuntil(ks, KS_SEP_LINE, &str, &dret) >= 0) {
        char *p, *q;
        int tid, beg, end, pos;
        bam_mplp_t mplp;

        for (p = q = str.s; *p && *p != '\t'; ++p);
        if (*p != '\t') goto bed_error;
        *p = 0; tid = bam_name2id(aux[0]->header, q); *p = '\t';
        if (tid < 0) goto bed_error;
        for (q = p = p + 1; isdigit(*p); ++p);
        if (*p != '\t') goto bed_error;
        *p = 0; beg = atoi(q); *p = '\t';
        for (q = p = p + 1; isdigit(*p); ++p);
        if (*p == '\t' || *p == 0) {
            int c = *p;
            *p = 0; end = atoi(q); *p = c;
        } else goto bed_error;

        for (i = 0; i < n; ++i) {
            if (aux[i]->iter) hts_itr_destroy(aux[i]->iter);
            aux[i]->iter = sam_itr_queryi(idx[i], tid, beg, end);
        }
        mplp = bam_mplp_init(n, read_bam, (void**)aux);
        bam_mplp_set_maxcnt(mplp, 64000);
        memset(cnt, 0, 8 * n);
        while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0)
            if (pos >= beg && pos < end)
                for (i = 0; i < n; ++i) cnt[i] += n_plp[i];
        for (i = 0; i < n; ++i) {
            kputc('\t', &str);
            kputl(cnt[i], &str);
        }
        fputs(str.s, pysam_stdout) & fputc('\n', pysam_stdout);
        bam_mplp_destroy(mplp);
        continue;

bed_error:
        fprintf(pysam_stderr, "Errors in BED line '%s'\n", str.s);
    }
    free(n_plp); free(plp);
    ks_destroy(ks);
    gzclose(fp);

    free(cnt);
    for (i = 0; i < n; ++i) {
        if (aux[i]->iter) hts_itr_destroy(aux[i]->iter);
        hts_idx_destroy(idx[i]);
        bam_hdr_destroy(aux[i]->header);
        sam_close(aux[i]->fp);
        free(aux[i]);
    }
    free(aux); free(idx);
    free(str.s);
    sam_global_args_free(&ga);
    return 0;
}
Пример #4
0
int main_bedcov(int argc, char *argv[])
{
    gzFile fp;
    kstring_t str;
    kstream_t *ks;
    hts_idx_t **idx;
    aux_t **aux;
    int *n_plp, dret, i, n, c, min_mapQ = 0;
    int64_t *cnt;
    const bam_pileup1_t **plp;

    while ((c = getopt(argc, argv, "Q:")) >= 0) {
        switch (c) {
        case 'Q': min_mapQ = atoi(optarg); break;
        }
    }
    if (optind + 2 > argc) {
        fprintf(stderr, "Usage: samtools bedcov <in.bed> <in1.bam> [...]\n");
        return 1;
    }
    memset(&str, 0, sizeof(kstring_t));
    n = argc - optind - 1;
    aux = calloc(n, sizeof(aux_t*));
    idx = calloc(n, sizeof(hts_idx_t*));
    for (i = 0; i < n; ++i) {
        aux[i] = calloc(1, sizeof(aux_t));
        aux[i]->min_mapQ = min_mapQ;
        aux[i]->fp = sam_open(argv[i+optind+1], "r");
        idx[i] = sam_index_load(aux[i]->fp, argv[i+optind+1]);
        if (aux[i]->fp == 0 || idx[i] == 0) {
            fprintf(stderr, "ERROR: fail to open index BAM file '%s'\n", argv[i+optind+1]);
            return 2;
        }
        // TODO bgzf_set_cache_size(aux[i]->fp, 20);
        aux[i]->header = sam_hdr_read(aux[i]->fp);
    }
    cnt = calloc(n, 8);

    fp = gzopen(argv[optind], "rb");
    ks = ks_init(fp);
    n_plp = calloc(n, sizeof(int));
    plp = calloc(n, sizeof(bam_pileup1_t*));
    while (ks_getuntil(ks, KS_SEP_LINE, &str, &dret) >= 0) {
        char *p, *q;
        int tid, beg, end, pos;
        bam_mplp_t mplp;

        for (p = q = str.s; *p && *p != '\t'; ++p);
        if (*p != '\t') goto bed_error;
        *p = 0; tid = bam_name2id(aux[0]->header, q); *p = '\t';
        if (tid < 0) goto bed_error;
        for (q = p = p + 1; isdigit(*p); ++p);
        if (*p != '\t') goto bed_error;
        *p = 0; beg = atoi(q); *p = '\t';
        for (q = p = p + 1; isdigit(*p); ++p);
        if (*p == '\t' || *p == 0) {
            int c = *p;
            *p = 0; end = atoi(q); *p = c;
        } else goto bed_error;

        for (i = 0; i < n; ++i) {
            if (aux[i]->iter) hts_itr_destroy(aux[i]->iter);
            aux[i]->iter = sam_itr_queryi(idx[i], tid, beg, end);
        }
        mplp = bam_mplp_init(n, read_bam, (void**)aux);
        bam_mplp_set_maxcnt(mplp, 64000);
        memset(cnt, 0, 8 * n);
        while (bam_mplp_auto(mplp, &tid, &pos, n_plp, plp) > 0)
            if (pos >= beg && pos < end)
                for (i = 0; i < n; ++i) cnt[i] += n_plp[i];
        for (i = 0; i < n; ++i) {
            kputc('\t', &str);
            kputl(cnt[i], &str);
        }
        puts(str.s);
        bam_mplp_destroy(mplp);
        continue;

bed_error:
        fprintf(stderr, "Errors in BED line '%s'\n", str.s);
    }
    free(n_plp); free(plp);
    ks_destroy(ks);
    gzclose(fp);

    free(cnt);
    for (i = 0; i < n; ++i) {
        if (aux[i]->iter) hts_itr_destroy(aux[i]->iter);
        hts_idx_destroy(idx[i]);
        bam_hdr_destroy(aux[i]->header);
        sam_close(aux[i]->fp);
        free(aux[i]);
    }
    free(aux); free(idx);
    free(str.s);
    return 0;
}