SEXP rph_tree_scale(SEXP treeStr, SEXP scaleP, SEXP nodeStr,
SEXP includeLeadingP) {
TreeNode *tr = rph_tree_new(treeStr);
double scale = NUMERIC_VALUE(scaleP);
char *newTreeStr;
SEXP result;
if (nodeStr != R_NilValue) {
TreeNode *n;
int includeLeading=LOGICAL_VALUE(includeLeadingP);
n = tr_get_node(tr, CHARACTER_VALUE(nodeStr));
if (n == NULL) {
tr_name_ancestors(tr);
n = tr_get_node(tr, CHARACTER_VALUE(nodeStr));
if (n == NULL)
die("No node named %s in %s\n", CHARACTER_VALUE(nodeStr),
CHARACTER_VALUE(treeStr));
}
tr_scale_subtree(tr, n, scale, includeLeading);
}
else tr_scale(tr, scale);
newTreeStr = tr_to_string(tr, 1);
PROTECT(result = NEW_CHARACTER(1));
SET_STRING_ELT(result, 0, mkChar(newTreeStr));
UNPROTECT(1);
return result;
}
SEXP rph_tree_name_ancestors(SEXP treeStr) {
TreeNode *tr = rph_tree_new(treeStr);
char *newTreeStr;
SEXP result;
tr_name_ancestors(tr);
newTreeStr = tr_to_string(tr, 1);
PROTECT(result = NEW_CHARACTER(1));
SET_STRING_ELT(result, 0, mkChar(newTreeStr));
UNPROTECT(1);
return result;
}
SEXP rph_tree_subtree(SEXP treeStr, SEXP nodeStr) {
TreeNode *tr = rph_tree_new(treeStr);
TreeNode *n;
char *newTreeStr;
SEXP result;
n = tr_get_node(tr, CHARACTER_VALUE(nodeStr));
if (n == NULL) {
tr_name_ancestors(tr);
n = tr_get_node(tr, CHARACTER_VALUE(nodeStr));
if (n == NULL)
die("No node named %s", CHARACTER_VALUE(nodeStr));
}
tr_prune_supertree(&tr, n);
newTreeStr = tr_to_string(tr, 1);
PROTECT(result = NEW_CHARACTER(1));
SET_STRING_ELT(result, 0, mkChar(newTreeStr));
UNPROTECT(1);
return result;
}
int main(int argc, char *argv[]) {
char c;
int opt_idx, node;
FILE *out_f = NULL, *msa_f, *mod_f;
char *out_root;
TreeModel *mod;
MSA *msa;
char out_fname[STR_MED_LEN];
struct option long_opts[] = {
{"refseq", 1, 0, 'r'},
{"msa-format", 1, 0, 'i'},
{"seqs", 1, 0, 's'},
{"exclude", 0, 0, 'x'},
{"no-probs", 0, 0, 'n'},
{"suff-stats", 0, 0, 'S'},
{"encode", 1, 0, 'e'},
{"keep-gaps", 0, 0, 'k'},
{"gibbs", 1, 0, 'G'},
{"help", 0, 0, 'h'},
{0, 0, 0, 0}
};
/* arguments and defaults for options */
FILE *refseq_f = NULL;
msa_format_type msa_format = UNKNOWN_FORMAT;
int suff_stats = FALSE, exclude = FALSE, keep_gaps = FALSE, do_probs = TRUE;
List *seqlist = NULL;
PbsCode *code = NULL;
int gibbs_nsamples = -1;
while ((c = (char)getopt_long(argc, argv, "r:i:s:e:knxSh", long_opts, &opt_idx)) != -1) {
switch (c) {
case 'r':
refseq_f = phast_fopen(optarg, "r");
break;
case 'i':
msa_format = msa_str_to_format(optarg);
if (msa_format == UNKNOWN_FORMAT)
die("ERROR: unrecognized alignment format.\n");
break;
case 'S':
suff_stats = TRUE;
break;
case 'e':
code = pbs_new_from_file(phast_fopen(optarg, "r"));
break;
case 's':
seqlist = get_arg_list(optarg);
break;
case 'x':
exclude = TRUE;
break;
case 'n':
do_probs = FALSE;
break;
case 'k':
keep_gaps = TRUE;
break;
case 'G':
gibbs_nsamples = get_arg_int_bounds(optarg, 1, INFTY);
break;
case 'h':
printf("%s", HELP);
exit(0);
case '?':
die("Bad argument. Try 'prequel -h'.\n");
}
}
if (optind != argc - 3)
die("Three arguments required. Try 'prequel -h'.\n");
set_seed(-1);
if (!do_probs && (suff_stats || code != NULL))
die("ERROR: --no-probs can't be used with --suff-stats or --encode.\n");
msa_f = phast_fopen(argv[optind], "r");
if (msa_format == UNKNOWN_FORMAT)
msa_format = msa_format_for_content(msa_f, 1);
fprintf(stderr, "Reading alignment from %s...\n", argv[optind]);
if (msa_format == MAF) {
msa = maf_read(msa_f, refseq_f, 1, NULL, NULL, NULL, -1, !suff_stats, NULL,
NO_STRIP, FALSE);
/* (no need to store order if suff_stats mode) */
}
else
msa = msa_new_from_file_define_format(msa_f, msa_format, NULL);
if (msa->ss == NULL) {
fprintf(stderr, "Extracting sufficient statistics...\n");
ss_from_msas(msa, 1, TRUE, NULL, NULL, NULL, -1, 0);
}
else if (msa->ss->tuple_idx == NULL && !suff_stats)
die("ERROR: ordered representation of alignment required unless --suff-stats.\n");
mod_f = phast_fopen(argv[optind+1], "r");
out_root = argv[optind+2];
mod = tm_new_from_file(mod_f, 1);
/* MH prune just like in phastcons */
int old_nnodes = mod->tree->nnodes;
List *pruned_names = lst_new_ptr(msa->nseqs);
tm_prune(mod, msa, pruned_names);
if (lst_size(pruned_names) == (old_nnodes + 1) / 2)
die("ERROR: no match for leaves of tree in alignment (leaf names must match alignment names).\n");
if (lst_size(pruned_names) > 0) {
fprintf(stderr, "WARNING: pruned away leaves of tree with no match in alignment (");
int j;
for (j = 0; j < lst_size(pruned_names); j++)
fprintf(stderr, "%s%s", ((String*)lst_get_ptr(pruned_names, j))->chars, j < lst_size(pruned_names) - 1 ? ", " : ").\n");
}
lst_free_strings(pruned_names);
tr_name_ancestors(mod->tree);
if (mod->order != 0)
die("ERROR: Only single nucleotide models are supported.\n");
if (mod->nratecats > 1)
die("ERROR: Rate variation not supported.\n");
mod->tree_posteriors = tl_new_tree_posteriors(mod, msa, TRUE, FALSE,
FALSE, FALSE, FALSE, FALSE,
FALSE);
fprintf(stderr, "Computing posterior probabilities...\n");
if (gibbs_nsamples > 0)
die("ERROR: --gibbs not implemented yet.");
/* gb_sample_ancestral_seqs(mod, msa, mod->tree_posteriors, gibbs_nsamples); */
else
tl_compute_log_likelihood(mod, msa, NULL, NULL, -1, mod->tree_posteriors);
fprintf(stderr, "Reconstructing indels by parsimony...\n");
do_indels(msa, mod);
for (node = 0; node < mod->tree->nnodes; node++) {
int i, j;
TreeNode *n = lst_get_ptr(mod->tree->nodes, node);
if (n->lchild == NULL || n->rchild == NULL) continue;
if (seqlist != NULL) {
int in_list = str_in_list_charstr(n->name, seqlist);
if ((in_list && exclude) || (!in_list && !exclude))
continue;
}
fprintf(stderr, "Writing output for ancestral node '%s'...\n",
n->name);
if (suff_stats) {
if (out_f == NULL) {
sprintf(out_fname, "%s.stats", out_root);
out_f = phast_fopen(out_fname, "w+");
fprintf(out_f, "#count\t");
for (j = 0; j < mod->rate_matrix->size; j++)
fprintf(out_f, "p(%c)%c", mod->rate_matrix->states[j],
j == mod->rate_matrix->size - 1 ? '\n' : '\t');
}
for (i = 0; i < msa->ss->ntuples; i++) {
if (mod->tree_posteriors->base_probs[0][0][node][i] == -1)
continue; /* no base this node */
fprintf(out_f, "%.0f\t", msa->ss->counts[i]);
for (j = 0; j < mod->rate_matrix->size; j++) {
fprintf(out_f, "%f%c",
mod->tree_posteriors->base_probs[0][j][node][i],
j == mod->rate_matrix->size - 1 ? '\n' : '\t');
}
}
}
else if (code == NULL && do_probs) { /* ordinary sequence-by-sequence
output */
sprintf(out_fname, "%s.%s.probs", out_root, n->name);
out_f = phast_fopen(out_fname, "w+");
fprintf(out_f, "#");
for (j = 0; j < mod->rate_matrix->size; j++)
fprintf(out_f, "p(%c)%c", mod->rate_matrix->states[j],
j == mod->rate_matrix->size - 1 ? '\n' : '\t');
for (i = 0; i < msa->length; i++) {
if (mod->tree_posteriors->base_probs[0][0][node][msa->ss->tuple_idx[i]] == -1) {
/* no base */
if (keep_gaps) fprintf(out_f, "-\n");
/* otherwise do nothing */
}
else
for (j = 0; j < mod->rate_matrix->size; j++)
fprintf(out_f, "%f%c",
mod->tree_posteriors->base_probs[0][j][node][msa->ss->tuple_idx[i]],
j == mod->rate_matrix->size - 1 ? '\n' : '\t');
}
phast_fclose(out_f);
}
else if (code == NULL && !do_probs) { /* write point estimates
to FASTA file */
char *outseq = smalloc((msa->length + 1) * sizeof(char));
int len = 0;
for (i = 0; i < msa->length; i++) {
if (mod->tree_posteriors->base_probs[0][0][node][msa->ss->tuple_idx[i]] == -1) {
/* no base */
if (keep_gaps) outseq[len++] = GAP_CHAR;
/* otherwise do nothing */
}
else {
double maxprob = 0;
int maxidx = -1;
for (j = 0; j < mod->rate_matrix->size; j++) {
if (mod->tree_posteriors->base_probs[0][j][node][msa->ss->tuple_idx[i]] > maxprob) {
maxprob = mod->tree_posteriors->base_probs[0][j][node][msa->ss->tuple_idx[i]];
maxidx = j;
}
}
outseq[len++] = mod->rate_matrix->states[maxidx];
}
}
outseq[len] = '\0';
/* print in FASTA format */
sprintf(out_fname, "%s.%s.fa", out_root, n->name);
out_f = phast_fopen(out_fname, "w+");
print_seq_fasta(out_f, outseq, n->name, len);
phast_fclose(out_f);
sfree(outseq);
}
else { /* encoded sequence-by-sequence
output */
double error, tot_error = 0;
int ngaps = 0;
Vector *v;
unsigned *encoded;
/* first encode tuple by tuple */
v = vec_new(mod->rate_matrix->size);
encoded = smalloc(msa->ss->ntuples * sizeof(unsigned));
for (i = 0; i < msa->ss->ntuples; i++) {
if (mod->tree_posteriors->base_probs[0][0][node][i] == -1) {
encoded[i] = code->gap_code;
ngaps += msa->ss->counts[i];
}
else {
for (j = 0; j < mod->rate_matrix->size; j++)
vec_set(v, j, mod->tree_posteriors->base_probs[0][j][node][i]);
encoded[i] = pbs_get_index(code, v, &error);
tot_error += error * msa->ss->counts[i];
}
}
vec_free(v);
/* now write site by site */
sprintf(out_fname, "%s.%s.bin", out_root, n->name);
out_f = phast_fopen(out_fname, "w+");
for (i = 0; i < msa->length; i++) {
if (keep_gaps || encoded[msa->ss->tuple_idx[i]] != code->gap_code)
pbs_write_binary(code, encoded[msa->ss->tuple_idx[i]], out_f);
}
fprintf(stderr, "Average approximation error ('%s'): %f bits\n",
n->name, tot_error/(msa->length - ngaps));
sfree(encoded);
}
}
fprintf(stderr, "Done.\n");
return 0;
}
int main(int argc, char *argv[]) {
List *pruned_names = lst_new_ptr(5);
TreeModel *source_mod;
MSA *msa = NULL, *out_msa;
IndelHistory *ih;
char *read_hist_fname = NULL;
char c;
int opt_idx, old_nnodes, i;
msa_format_type msa_format = UNKNOWN_FORMAT;
int output_alignment = FALSE, ia_names = FALSE;
struct option long_opts[] = {
{"msa-format", 1, 0, 'i'},
{"output-alignment", 0, 0, 'A'},
{"read-history", 1, 0, 'H'},
{"ia-names", 0, 0, 'I'},
{"help", 0, 0, 'h'},
{0, 0, 0, 0}
};
while ((c = getopt_long(argc, argv, "i:H:AIh", long_opts, &opt_idx)) != -1) {
switch (c) {
case 'i':
msa_format = msa_str_to_format(optarg);
if (msa_format == -1)
die("ERROR: unrecognized alignment format.\n");
break;
case 'A':
output_alignment = TRUE;
break;
case 'H':
read_hist_fname = optarg;
break;
case 'I':
ia_names = TRUE;
break;
case 'h':
printf("%s", HELP);
exit(0);
case '?':
die("Bad argument. Try 'indelHistory -h'.\n");
}
}
set_seed(-1);
if (read_hist_fname != NULL) {
fprintf(stderr, "Reading indel history from %s...\n", read_hist_fname);
ih = ih_new_from_file(phast_fopen(read_hist_fname, "r"));
}
else {
FILE *mfile;
if (optind != argc - 2)
die("Two arguments required. Try 'indelHistory -h'.\n");
fprintf(stderr, "Reading alignment from %s...\n", argv[optind]);
mfile = phast_fopen(argv[optind], "r");
if (msa_format == UNKNOWN_FORMAT)
msa_format = msa_format_for_content(mfile, 1);
msa = msa_new_from_file_define_format(mfile, msa_format, "ACGTNB^.-");
phast_fclose(mfile);
if (msa->seqs == NULL && (msa->ss == NULL || msa->ss->tuple_idx == NULL))
die("ERROR: ordered representation of alignment required.\n");
fprintf(stderr, "Reading tree from %s...\n", argv[optind+1]);
source_mod = tm_new_from_file(phast_fopen(argv[optind+1], "r"), 1);
/* prune tree, if necessary */
old_nnodes = source_mod->tree->nnodes;
tm_prune(source_mod, msa, pruned_names);
if (lst_size(pruned_names) == (old_nnodes + 1) / 2)
die("ERROR: no match for leaves of tree in alignment (leaf names must match alignment names).\n");
if (lst_size(pruned_names) > 0) {
fprintf(stderr, "WARNING: pruned away leaves of tree with no match in alignment (");
for (i = 0; i < lst_size(pruned_names); i++)
fprintf(stderr, "%s%s", ((String*)lst_get_ptr(pruned_names, i))->chars,
i < lst_size(pruned_names) - 1 ? ", " : ").\n");
}
lst_free(pruned_names);
tr_name_ancestors(source_mod->tree);
if (msa->nseqs > (source_mod->tree->nnodes + 1) / 2) { /* assume ancestral
seqs specified
in this case */
if (ia_names) {
fprintf(stderr, "Converting sequence names...\n");
ih_convert_ia_names(msa, source_mod->tree);
}
fprintf(stderr, "Extracting indel history from alignment...\n");
ih = ih_extract_from_alignment(msa, source_mod->tree);
}
else { /* infer by parsimony */
if (msa->ss == NULL) {
fprintf(stderr, "Extracting sufficient statistics...\n");
ss_from_msas(msa, 1, TRUE, NULL, NULL, NULL, -1, 0);
}
fprintf(stderr, "Inferring indel history by parsimony...\n");
ih = ih_reconstruct(msa, source_mod->tree);
}
}
if (output_alignment) {
out_msa = ih_as_alignment(ih, msa);
msa_print(stdout, out_msa, FASTA, FALSE);
}
else
ih_print(ih, stdout,
read_hist_fname != NULL ? read_hist_fname : argv[optind],
"indelHistory");
fprintf(stderr, "Done.\n");
return 0;
}
int main(int argc, char *argv[]) {
char c;
char *msa_fname = NULL;
int opt_idx, i, old_nnodes;
MSA *msa;
List *pruned_names = lst_new_ptr(5), *tmpl;
BDPhyloHmm *bdphmm;
GFF_Set *predictions;
int found = FALSE;
List *ignore_types = lst_new_ptr(1);
struct option long_opts[] = {
{"refseq", 1, 0, 'M'},
{"msa-format", 1, 0, 'i'},
{"refidx", 1, 0, 'r'},
{"rho", 1, 0, 'R'},
{"phi", 1, 0, 'p'},
{"transitions", 1, 0, 't'},
{"expected-length", 1, 0, 'E'},
{"target-coverage", 1, 0, 'C'},
{"seqname", 1, 0, 'N'},
{"idpref", 1, 0, 'P'},
{"indel-model", 1, 0, 'I'},
{"indel-history", 1, 0, 'H'},
{"help", 0, 0, 'h'},
{0, 0, 0, 0}
};
/* arguments and defaults for options */
FILE *refseq_f = NULL, *msa_f = NULL;
msa_format_type msa_format = UNKNOWN_FORMAT;
TreeModel *source_mod;
double rho = DEFAULT_RHO, mu = DEFAULT_MU, nu = DEFAULT_NU,
phi = DEFAULT_PHI, gamma = -1, omega = -1,
alpha_c = -1, beta_c = -1, tau_c = -1,
alpha_n = -1, beta_n = -1, tau_n = -1;
int set_transitions = FALSE, refidx = 1, estim_phi = TRUE,
estim_gamma = TRUE, estim_omega = TRUE;
char *seqname = NULL, *idpref = NULL;
IndelHistory *ih = NULL;
while ((c = getopt_long(argc, argv, "R:t:p:E:C:r:M:i:N:P:I:H:h", long_opts, &opt_idx)) != -1) {
switch (c) {
case 'R':
rho = get_arg_dbl_bounds(optarg, 0, 1);
break;
case 't':
if (optarg[0] != '~') estim_gamma = estim_omega = FALSE;
else optarg = &optarg[1];
set_transitions = TRUE;
tmpl = get_arg_list_dbl(optarg);
if (lst_size(tmpl) != 2)
die("ERROR: bad argument to --transitions.\n");
mu = lst_get_dbl(tmpl, 0);
nu = lst_get_dbl(tmpl, 1);
if (mu <= 0 || mu >= 1 || nu <= 0 || nu >= 1)
die("ERROR: bad argument to --transitions.\n");
lst_free(tmpl);
break;
case 'p':
if (optarg[0] != '~') estim_phi = FALSE;
else optarg = &optarg[1];
phi = get_arg_dbl_bounds(optarg, 0, 1);
break;
case 'E':
if (optarg[0] != '~') estim_omega = FALSE;
else optarg = &optarg[1];
omega = get_arg_dbl_bounds(optarg, 1, INFTY);
mu = 1/omega;
break;
case 'C':
if (optarg[0] != '~') estim_gamma = FALSE;
else optarg = &optarg[1];
gamma = get_arg_dbl_bounds(optarg, 0, 1);
break;
case 'r':
refidx = get_arg_int_bounds(optarg, 0, INFTY);
break;
case 'M':
refseq_f = phast_fopen(optarg, "r");
break;
case 'i':
msa_format = msa_str_to_format(optarg);
if (msa_format == UNKNOWN_FORMAT)
die("ERROR: unrecognized alignment format.\n");
break;
case 'N':
seqname = optarg;
break;
case 'P':
idpref = optarg;
break;
case 'I':
tmpl = get_arg_list_dbl(optarg);
if (lst_size(tmpl) != 3 && lst_size(tmpl) != 6)
die("ERROR: bad argument to --indel-model.\n");
alpha_n = lst_get_dbl(tmpl, 0);
beta_n = lst_get_dbl(tmpl, 1);
tau_n = lst_get_dbl(tmpl, 2);
if (lst_size(tmpl) == 6) {
alpha_c = lst_get_dbl(tmpl, 3);
beta_c = lst_get_dbl(tmpl, 4);
tau_c = lst_get_dbl(tmpl, 5);
}
else {
alpha_c = alpha_n; beta_c = beta_n; tau_c = tau_n;
}
if (alpha_c <= 0 || alpha_c >= 1 || beta_c <= 0 || beta_c >= 1 ||
tau_c <= 0 || tau_c >= 1 || alpha_n <= 0 || alpha_n >= 1 ||
beta_n <= 0 || beta_n >= 1 || tau_n <= 0 || tau_n >= 1)
die("ERROR: bad argument to --indel-model.\n");
break;
case 'H':
fprintf(stderr, "Reading indel history from %s...\n", optarg);
ih = ih_new_from_file(phast_fopen(optarg, "r"));
break;
case 'h':
printf("%s", HELP);
exit(0);
case '?':
die("Bad argument. Try 'dless -h'.\n");
}
}
if (optind != argc - 1)
die("Missing alignment file or model file. Try 'dless -h'.\n");
if (set_transitions && (gamma != -1 || omega != -1))
die("ERROR: --transitions and --target-coverage/--expected-length cannot be used together.\n");
if ((gamma != -1 && omega == -1) || (gamma == -1 && omega != -1))
die("ERROR: --target-coverage and --expecteed-length must be used together.\n");
set_seed(-1);
if (gamma != -1)
nu = gamma/(1-gamma) * mu;
fprintf(stderr, "Reading tree model from %s...\n", argv[optind]);
source_mod = tm_new_from_file(phast_fopen(argv[optind], "r"), 1);
if (source_mod->nratecats > 1)
die("ERROR: rate variation not currently supported.\n");
if (source_mod->order > 0)
die("ERROR: only single nucleotide models are currently supported.\n");
if (!tm_is_reversible(source_mod))
phast_warning("WARNING: p-value computation assumes reversibility and your model is non-reversible.\n");
/* read alignment */
msa_f = phast_fopen(argv[optind], "r");
fprintf(stderr, "Reading alignment from %s...\n", argv[optind]);
if (msa_format == UNKNOWN_FORMAT)
msa_format = msa_format_for_content(msa_f, 1);
if (msa_format == MAF) {
msa = maf_read(msa_f, refseq_f, 1, NULL, NULL, NULL, -1, TRUE, NULL,
NO_STRIP, FALSE);
}
else
msa = msa_new_from_file_define_format(msa_f, msa_format, NULL);
if (msa_alph_has_lowercase(msa)) msa_toupper(msa);
msa_remove_N_from_alph(msa);
if (msa->ss == NULL) {
fprintf(stderr, "Extracting sufficient statistics...\n");
ss_from_msas(msa, 1, TRUE, NULL, NULL, NULL, -1, 0);
}
else if (msa->ss->tuple_idx == NULL)
die("ERROR: ordered representation of alignment required unless --suff-stats.\n");
/* prune tree, if necessary */
old_nnodes = source_mod->tree->nnodes;
tm_prune(source_mod, msa, pruned_names);
if (lst_size(pruned_names) == (old_nnodes + 1) / 2)
die("ERROR: no match for leaves of tree in alignment (leaf names must match alignment names).\n");
if (lst_size(pruned_names) > 0) {
fprintf(stderr, "WARNING: pruned away leaves of tree with no match in alignment (");
for (i = 0; i < lst_size(pruned_names); i++)
fprintf(stderr, "%s%s", ((String*)lst_get_ptr(pruned_names, i))->chars,
i < lst_size(pruned_names) - 1 ? ", " : ").\n");
}
/* this has to be done after pruning tree */
tr_name_ancestors(source_mod->tree);
/* also make sure match for reference sequence in tree */
if (refidx > 0) {
for (i = 0, found = FALSE; !found && i < source_mod->tree->nnodes; i++) {
TreeNode *n = lst_get_ptr(source_mod->tree->nodes, i);
if (!strcmp(n->name, msa->names[refidx-1]))
found = TRUE;
}
if (!found) die("ERROR: no match for reference sequence in tree.\n");
}
/* checks for indel model */
if (alpha_c > 0) {
if (ih == NULL) {
fprintf(stderr, "Reconstructing indel history by parsimony...\n");
ih = ih_reconstruct(msa, source_mod->tree);
}
else {
if (ih->ncols != msa->length)
die("ERROR: indel history doesn't seem to match alignment.\n");
if (ih->tree->nnodes != source_mod->tree->nnodes)
die("ERROR: indel history doesn't seem to match tree model.\n");
}
}
bdphmm = bd_new(source_mod, rho, mu, nu, phi, alpha_c, beta_c, tau_c,
alpha_n, beta_n, tau_n, estim_gamma, estim_omega,
estim_phi);
/* compute emissions */
phmm_compute_emissions(bdphmm->phmm, msa, FALSE);
/* add emissions for indel model, if necessary */
if (alpha_c > 0) {
fprintf(stderr, "Adjusting emissions for indels...\n");
bd_add_indel_emissions(bdphmm, ih);
}
/* postprocess for missing data (requires special handling) */
fprintf(stderr, "Adjusting emissions for missing data...\n");
bd_handle_missing_data(bdphmm, msa);
if (estim_gamma || estim_omega || estim_phi) {
fprintf(stderr, "Estimating free parameters...\n");
bd_estimate_transitions(bdphmm, msa);
}
/* set seqname and idpref, if necessary */
if (seqname == NULL || idpref == NULL) {
/* derive default from file name root */
String *tmp = str_new_charstr(msa_fname);
if (!str_equals_charstr(tmp, "-")) {
str_remove_path(tmp);
str_root(tmp, '.');
if (idpref == NULL) idpref = copy_charstr(tmp->chars);
str_root(tmp, '.'); /* apply one more time for double suffix */
if (seqname == NULL) seqname = tmp->chars;
}
else if (seqname == NULL) seqname = "refseq";
}
/* obtain predictions */
fprintf(stderr, "Running Viterbi algorithm...\n");
predictions = phmm_predict_viterbi(bdphmm->phmm, seqname, NULL, idpref, NULL);
lst_push_ptr(ignore_types, str_new_charstr("nonconserved"));
gff_filter_by_type(predictions, ignore_types, TRUE, NULL);
/* score predictions */
fprintf(stderr, "Scoring predictions...\n");
bd_score_predictions(bdphmm, predictions);
/* can free emissions now */
for (i = 0; i < bdphmm->phmm->hmm->nstates; i++)
sfree(bdphmm->phmm->emissions[i]);
sfree(bdphmm->phmm->emissions);
bdphmm->phmm->emissions = NULL;
/* convert GFF to coord frame of reference sequence and adjust
coords by idx_offset, if necessary */
if (refidx != 0 || msa->idx_offset != 0)
msa_map_gff_coords(msa, predictions, 0, refidx, msa->idx_offset);
if (refidx != 0)
gff_flatten(predictions);
/* necessary because coord conversion might create overlapping
features (can happen in deletions in reference sequence) */
/* now output predictions */
fprintf(stderr, "Writing GFF to stdout...\n");
gff_print_set(stdout, predictions);
fprintf(stderr, "Done.\n");
return 0;
}
int main(int argc, char *argv[]) {
/* variables for options, with defaults */
TreeNode *tree = NULL, *merge_tree = NULL, *extrapolate_tree = NULL;
Hashtable *rename_hash = NULL;
double scale_factor = 1;
List *prune_names = NULL, *label = NULL, *labelType = NULL;
int prune_all_but = FALSE, tree_only = FALSE, dissect = FALSE,
name_ancestors = FALSE, with_branch = FALSE, print_branchlen=FALSE,
inNewick=FALSE, no_branchlen = FALSE, print_distance_to_root = FALSE;
TreeModel *mod = NULL, *merge_mod = NULL;
char *reroot_name = NULL, *subtree_name =NULL, *get_subtree_name = NULL,
*node_distance_name = NULL;
/* other variables */
String *suffix, *optstr;
char c;
int i, opt_idx;
TreeNode *n;
struct option long_opts[] = {
{"scale", 1, 0, 's'},
{"extrapolate", 1, 0, 'e'},
{"prune", 1, 0, 'p'},
{"prune-all-but", 1, 0, 'P'},
{"get-subtree", 1, 0, 'g'},
{"merge", 1, 0, 'm'},
{"rename", 1, 0, 'r'},
{"tree-only", 0, 0, 't'},
{"no-branchlen", 0, 0, 'N'},
{"dissect", 0, 0, 'd'},
{"name-ancestors", 0, 0, 'a'},
{"reroot", 1, 0, 'R'},
{"with-branch", 1, 0, 'B'},
{"subtree", 1, 0, 'S'},
{"branchlen", 0, 0, 'b'},
{"newick", 0, 0, 'n'},
{"label-subtree", 1, 0, 'L'},
{"label-branches", 1, 0, 'l'},
{"help", 0, 0, 'h'},
{0, 0, 0, 0}
};
while ((c = getopt_long(argc, argv, "s:p:P:g:m:r:R:B:S:D:l:L:adtNbnh",
long_opts, &opt_idx)) != -1) {
switch (c) {
case 's':
scale_factor = get_arg_dbl_bounds(optarg, 0, INFTY);
break;
case 'e':
if (!strcmp(optarg, "default")) {
optarg = smalloc(1000 * sizeof(char));
#if defined(__MINGW32__)
sprintf(optarg, "%s\\data\\exoniphy\\mammals\\cftr25_hybrid.nh",
PHAST_HOME);
#else
sprintf(optarg, "%s/data/exoniphy/mammals/cftr25_hybrid.nh",
PHAST_HOME);
#endif
}
extrapolate_tree = tr_new_from_file(phast_fopen(optarg, "r"));
break;
case 'p':
prune_names = get_arg_list(optarg);
break;
case 'P':
prune_names = get_arg_list(optarg);
prune_all_but = TRUE;
break;
case 'g':
get_subtree_name = optarg;
break;
case 'm':
suffix = str_new_charstr(optarg);
str_suffix(suffix, '.');
if (str_equals_charstr(suffix, "nh"))
merge_tree = tr_new_from_file(phast_fopen(optarg, "r"));
else {
merge_mod = tm_new_from_file(phast_fopen(optarg, "r"), 1);
merge_tree = merge_mod->tree;
}
break;
case 'r':
rename_hash = make_name_hash(optarg);
break;
case 't':
tree_only = TRUE;
break;
case 'N':
no_branchlen = TRUE;
tree_only = TRUE;
break;
case 'd':
dissect = TRUE;
break;
case 'b':
print_branchlen = TRUE;
break;
case 'D':
print_distance_to_root = TRUE;
node_distance_name = optarg;
break;
case 'R':
reroot_name = optarg;
break;
case 'B':
with_branch = TRUE;
break;
case 'a':
name_ancestors = TRUE;
break;
case 'S':
subtree_name = optarg;
break;
case 'n':
inNewick=TRUE;
break;
case 'L': //do the same for --label--subtree and --label-branches
case 'l':
if (label == NULL) {
label = lst_new_ptr(1);
labelType = lst_new_int(1);
}
optstr = str_new_charstr(optarg);
lst_push_ptr(label, optstr);
lst_push_int(labelType, (int)c);
break;
case 'h':
usage(argv[0]);
case '?':
die("Bad argument. Try '%s -h'.\n", argv[0]);
}
}
if (optind != argc - 1)
die("Input filename required. Try '%s -h'.\n", argv[0]);
if (merge_tree != NULL && extrapolate_tree != NULL)
die("ERROR: Can't use --merge and --extrapolate together");
set_seed(-1);
suffix = str_new_charstr(argv[optind]);
str_suffix(suffix, '.');
if (inNewick || str_equals_charstr(suffix, "nh")) {
tree = tr_new_from_file(phast_fopen(argv[optind], "r"));
tree_only = TRUE; /* can't output tree model in this case */
}
else {
mod = tm_new_from_file(phast_fopen(argv[optind], "r"), 1);
tree = mod->tree;
}
if (prune_names != NULL) {
tr_prune(&tree, prune_names, prune_all_but, NULL);
if (mod != NULL) mod->tree = tree; /* root may have changed */
}
if (get_subtree_name != NULL) {
n = tr_get_node(tree, get_subtree_name);
if (n == NULL) {
tr_name_ancestors(tree);
n = tr_get_node(tree, get_subtree_name);
if (n == NULL) {
die("ERROR: no node named '%s'.\n", subtree_name);
}
}
tr_prune_supertree(&tree, n);
if (mod != NULL) mod->tree = tree;
}
if (merge_tree != NULL) {
tree = tr_hybrid(tree, merge_tree);
if (mod != NULL) mod->tree = tree;
}
else if (extrapolate_tree != NULL) {
tr_scale_by_subtree(extrapolate_tree, tree);
tree = extrapolate_tree;
if (mod != NULL) mod->tree = tree;
}
if (scale_factor != 1) {
if (subtree_name == NULL)
tr_scale(tree, scale_factor);
else {
n = tr_get_node(tree, subtree_name);
if (n == NULL) die("ERROR: no node named '%s'.\n", subtree_name);
tr_scale_subtree(tree, n, scale_factor, with_branch);
}
}
if (name_ancestors)
tr_name_ancestors(tree);
if (rename_hash != NULL) {
char *newname;
for (i = 0; i < tree->nnodes; i++) {
n = lst_get_ptr(tree->nodes, i);
if (n->name != NULL && n->name[0] != '\0' &&
(newname = hsh_get(rename_hash, n->name)) != (char*)-1) {
strcpy(n->name, newname);
}
}
}
if (reroot_name != NULL) {
n = tr_get_node(tree, reroot_name);
if (n == NULL) die("ERROR: no node named '%s'.\n", reroot_name);
tr_reroot(tree, n, with_branch);
if (mod != NULL) mod->tree = with_branch ? n->parent : n;
tree = with_branch ? n->parent : n;
}
if (label != NULL) {
for (i=0; i < lst_size(label); i++) {
String *currstr = (String*)lst_get_ptr(label, i), *arg1, *labelVal;
List *tmplst = lst_new_ptr(10);
String *nodename;
int j;
str_split(currstr, ":", tmplst);
if (lst_size(tmplst) != 2)
die("ERROR: bad argument to --label-branches or --label-subtree.\n");
arg1 = lst_get_ptr(tmplst, 0);
labelVal = lst_get_ptr(tmplst, 1);
lst_clear(tmplst);
if (lst_get_int(labelType, i) == (int)'l') {
str_split(arg1, ",", tmplst);
for (j=0; j < lst_size(tmplst); j++) {
nodename = (String*)lst_get_ptr(tmplst, j);
tr_label_node(tree, nodename->chars, labelVal->chars);
}
lst_free_strings(tmplst);
} else if (lst_get_int(labelType, i) == (int)'L') {
int include_leading_branch = FALSE;
TreeNode *node;
nodename = arg1;
node = tr_get_node(tree, nodename->chars);
if (node == NULL && nodename->chars[nodename->length-1] == '+') {
nodename->chars[--nodename->length] = '\0';
node = tr_get_node(tree, nodename->chars);
include_leading_branch = TRUE;
}
tr_label_subtree(tree, nodename->chars, include_leading_branch,
labelVal->chars);
} else die("ERROR got label_type %c\n", lst_get_int(labelType, (char)i));
str_free(arg1);
str_free(labelVal);
lst_free(tmplst);
str_free(currstr);
}
lst_free(label);
lst_free(labelType);
}
if (dissect)
tr_print_nodes(stdout, tree);
if (print_branchlen)
printf("TOTAL_TREE_LEN: %f\n", tr_total_len(tree));
if (print_distance_to_root) {
TreeNode *node = tr_get_node(tree, node_distance_name);
if (node == NULL)
die("ERROR: no node named '%s'.\n", node_distance_name);
printf("length(root-%s): %f\n", node_distance_name,
tr_distance_to_root(node));
}
if (dissect==0 && print_branchlen==0 && print_distance_to_root==0) {
if (tree_only)
tr_print(stdout, tree, no_branchlen==FALSE);
else
tm_print(stdout, mod);
}
return 0;
}
