int main(int argc, char *argv[])
/* Process command line. */
{
if (argc != 6)
errAbort("bad running of bwmake");
struct metaBig *mb = metaBigOpen(argv[1], NULL);
FILE *out = mustOpen(argv[2], "w");
int decimals = sqlSigned(argv[3]);
struct bed *section;
double na = NANUM;
boolean condense = FALSE;
enum wigOutType wot = get_wig_out_type(argv[4]);
if (sameString(argv[5], "condense"))
condense = TRUE;
for (section = mb->sections; section != NULL; section = section->next)
{
struct perBaseWig *pbw = perBaseWigLoadSingleContinue(mb, section->chrom, section->chromStart,
section->chromEnd, FALSE, na);
perBaseWigOutputNASkip(pbw, out, wot, decimals, NULL, FALSE, condense);
perBaseWigFree(&pbw);
}
carefulClose(&out);
metaBigClose(&mb);
return 0;
}
/* Load all the regions from a wig or bigWig into a list of arrays basically. */
struct perBaseWig* perBaseWigLoad(char* wigFile, char* chrom, int start, int end)
{
struct metaBig* mb = metaBigOpen(wigFile, NULL);
if (mb->type != isaBigWig) {
metaBigClose(&mb);
return NULL;
}
struct perBaseWig* list;
list = perBaseWigLoadContinue(mb, chrom, start, end);
metaBigClose(&mb);
return list;
}
struct perBaseWig* perBaseWigLoadSingle(char* wigFile, char* chrom, int start, int end, boolean reverse, double fill)
/* Load all the regions into one perBaseWig, but with gaps filled */
/* in with NA value */
{
struct metaBig* mb = metaBigOpen(wigFile, NULL);
if (mb->type != isaBigWig) {
metaBigClose(&mb);
return NULL;
}
struct perBaseWig* list;
list = perBaseWigLoadSingleContinue(mb, chrom, start, end, reverse, fill);
metaBigClose(&mb);
return list;
}
void bwtool_matrix(struct hash *options, char *favorites, char *regions, unsigned decimals,
double fill, char *range_s, char *bigfile, char *outputfile)
/* bwtool_matrix - main for matrix-creation program */
{
boolean do_k = (hashFindVal(options, "cluster") != NULL) ? TRUE : FALSE;
boolean do_tile = (hashFindVal(options, "tiled-averages") != NULL) ? TRUE : FALSE;
boolean keep_bed = (hashFindVal(options, "keep-bed") != NULL) ? TRUE : FALSE;
boolean starts = (hashFindVal(options, "starts") != NULL) ? TRUE : FALSE;
boolean ends = (hashFindVal(options, "ends") != NULL) ? TRUE : FALSE;
boolean lf_header = (hashFindVal(options, "long-form-header") != NULL) ? TRUE : FALSE;
char *centroid_file = (char *)hashFindVal(options, "cluster-centroids");
char *long_form = (char *)hashFindVal(options, "long-form");
boolean do_long_form = (long_form != NULL);
unsigned left = 0, right = 0;
int meta = 0;
int num_parse = parse_left_right(range_s, &left, &right, &meta);
boolean do_meta = (num_parse == 3);
int k = (int)sqlUnsigned((char *)hashOptionalVal(options, "cluster", "0"));
int tile = (int)sqlUnsigned((char *)hashOptionalVal(options, "tiled-averages", "1"));
if ((do_k) && ((k < 2) || (k > 10)))
errAbort("k should be between 2 and 10\n");
if ((do_tile) && (tile < 2))
errAbort("tiling should be done for larger regions");
if ((left % tile != 0) || (right % tile != 0))
errAbort("tiling should be multiple of both up and down values");
if (do_meta && starts && ends)
warn("meta uses -starts and -ends anyway");
else if ((do_meta) && (starts || ends))
warn("-starts and -ends both automatically used with meta");
if (do_meta && tile)
errAbort("meta not compatible with -tile... yet");
struct slName *bw_names = slNameListFromComma(bigfile);
struct slName *bw_name;
struct slName *labels_from_file = NULL;
int num_bigwigs = check_for_list_files(&bw_names, &labels_from_file);
struct slName *labels = setup_labels(long_form, bw_names, &labels_from_file);
struct bed6 *regs = NULL;
struct bed6 *regions_left = NULL, *regions_right = NULL, *regions_meta = NULL;
struct perBaseMatrix *pbm = NULL;
int i;
if (do_meta)
{
if (meta == -1)
{
meta = calculate_meta_file(regions);
fprintf(stderr, "calculated meta = %d bases\n", meta);
}
regions_left = load_and_recalculate_coords(regions, left, 0, FALSE, TRUE, FALSE);
regions_right = load_and_recalculate_coords(regions, 0, right, FALSE, FALSE, TRUE);
regions_meta = (meta > 0) ? readBed6Soft(regions) : NULL;
}
else
regs = load_and_recalculate_coords(regions, left, right, FALSE, starts, ends);
for (bw_name = bw_names; bw_name != NULL; bw_name = bw_name->next)
{
struct metaBig *mb = metaBigOpen(bw_name->name, NULL);
if (do_meta)
{
struct perBaseMatrix *one_pbm = load_perBaseMatrix(mb, regions_left, fill);
struct perBaseMatrix *right_pbm = load_perBaseMatrix(mb, regions_right, fill);
if (meta > 0)
{
struct perBaseMatrix *meta_pbm = load_meta_perBaseMatrix(mb, regions_meta, meta, fill);
fuse_pbm(&one_pbm, &meta_pbm, TRUE);
}
fuse_pbm(&one_pbm, &right_pbm, TRUE);
fuse_pbm(&pbm, &one_pbm, FALSE);
}
else
{
struct perBaseMatrix *one_pbm = (do_tile) ? load_ave_perBaseMatrix(mb, regs, tile, fill) :
load_perBaseMatrix(mb, regs, fill);
fuse_pbm(&pbm, &one_pbm, FALSE);
}
metaBigClose(&mb);
}
if (do_k)
{
struct cluster_bed_matrix *cbm = NULL;
/* ordered by cluster with label in first column */
cbm = init_cbm_from_pbm(pbm, k);
do_kmeans_sort(cbm, 0.001, TRUE);
if (do_long_form)
output_cluster_matrix_long(cbm, labels, keep_bed, outputfile, lf_header);
else
output_cluster_matrix(cbm, decimals, keep_bed, outputfile);
if (centroid_file)
output_centroids(cbm, centroid_file, decimals);
free_cbm(&cbm);
}
else
{
if (do_long_form)
output_matrix_long(pbm, decimals, labels, keep_bed, left, right, tile, lf_header, outputfile);
else
output_matrix(pbm, decimals, keep_bed, outputfile);
/* unordered, no label */
free_perBaseMatrix(&pbm);
}
if (do_meta)
{
bed6FreeList(®ions_left);
bed6FreeList(®ions_right);
if (meta > 0)
bed6FreeList(®ions_meta);
}
else
bed6FreeList(®s);
}
void bwtool_window(struct hash *options, char *favorites, char *regions, unsigned decimals,
double fill, char *size_s, char *bigfile, char *output_file)
/* bwtool_window - main for the windowing program */
{
struct metaBig *mb = metaBigOpen(bigfile, regions);
boolean skip_na = (hashFindVal(options, "skip-NA") != NULL) ? TRUE : FALSE;
if (!isnan(fill) && skip_na)
errAbort("cannot use -skip_na with -fill");
boolean center = (hashFindVal(options, "center") != NULL) ? TRUE : FALSE;
int step = (int)sqlUnsigned((char *)hashOptionalVal(options, "step", "1"));
int size = sqlSigned(size_s);
if (size < 1)
errAbort("size must be >= 1 for bwtool window");
FILE *out = (output_file) ? mustOpen(output_file, "w") : stdout;
struct bed *section;
for (section = mb->sections; section != NULL; section = section->next)
{
if (size <= section->chromEnd - section->chromStart)
{
struct perBaseWig *pbw = perBaseWigLoadSingleContinue(mb, section->chrom, section->chromStart,
section->chromEnd, FALSE, fill);
int i, j;
for (i = 0; i <= pbw->len - size; i += step)
{
int s = pbw->chromStart + i;
int e = pbw->chromStart + i + size;
if (center)
{
s += size/2 - step/2;
e = s + step;
}
boolean has_NA = FALSE;
if (skip_na)
{
for (j = i; j < i + size; j++)
if (isnan(pbw->data[j]))
{
i = j-step+1;
has_NA = TRUE;
break;
}
}
if (!has_NA)
{
fprintf(out, "%s\t%d\t%d\t", pbw->chrom, s, e);
for (j = i; j < i + size; j++)
if (isnan(pbw->data[j]) && (j == i + size - 1))
fprintf(out, "NA\n");
else if (isnan(pbw->data[j]))
fprintf(out, "NA,");
else if (j == i + size - 1)
fprintf(out, "%0.*f\n", decimals, pbw->data[j]);
else
fprintf(out, "%0.*f,", decimals, pbw->data[j]);
}
}
perBaseWigFree(&pbw);
}
}
metaBigClose(&mb);
carefulClose(&out);
}
