void output_long_distance(long diffl, long tree1, long tree2, long trees_in_1,
long trees_in_2)
{
switch (tree_pairing) {
case ADJACENT_PAIRS:
if (output_scheme == VERBOSE ) {
fprintf (outfile, "Trees %ld and %ld: %ld\n", tree1, tree2, diffl);
} else if (output_scheme == SPARSE) {
fprintf (outfile, "%ld %ld %ld\n", tree1, tree2, diffl);
}
break;
case ALL_IN_FIRST:
if (output_scheme == VERBOSE) {
fprintf (outfile, "Trees %ld and %ld: %ld\n", tree1, tree2, diffl);
} else if (output_scheme == SPARSE) {
fprintf (outfile, "%ld %ld %ld\n", tree1, tree2, diffl );
} else if (output_scheme == FULL_MATRIX) {
output_matrix_long(diffl, tree1, tree2, trees_in_1, trees_in_2);
}
break;
case CORR_IN_1_AND_2:
if (output_scheme == VERBOSE) {
fprintf (outfile, "Tree pair %ld: %ld\n", tree1, diffl);
} else if (output_scheme == SPARSE) {
fprintf (outfile, "%ld %ld\n", tree1, diffl);
}
break;
case ALL_IN_1_AND_2:
if (output_scheme == VERBOSE)
fprintf (outfile, "Trees %ld and %ld: %ld\n", tree1, tree2, diffl);
else if (output_scheme == SPARSE)
fprintf (outfile, "%ld %ld %ld\n", tree1, tree2, diffl);
else if (output_scheme == FULL_MATRIX ) {
output_matrix_long(diffl, tree1, tree2, trees_in_1, trees_in_2);
}
break;
}
}
void bwtool_matrix(struct hash *options, char *favorites, char *regions, unsigned decimals,
double fill, char *range_s, char *bigfile, char *tmp_dir, 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 do_binary_matrix = (hashFindVal(options, "binary-matrix") != 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 && do_tile)
errAbort("meta not compatible with -tile... yet");
if (do_binary_matrix && do_long_form)
errAbort("Writing binary matrix is not compatible with -long-form... 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, 0);
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 = metaBigOpenWithTmpDir(bw_name->name, tmp_dir, 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
{
if (do_binary_matrix)
{
output_binary_cluster_matrix(cbm, keep_bed, outputfile);
}
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
{
if (do_binary_matrix)
{
output_binary_matrix(pbm, keep_bed, 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);
}
