static int gt_condenser_search_runner(GT_UNUSED int argc,
GT_UNUSED const char **argv,
GT_UNUSED int parsed_args,
void *tool_arguments,
GtError *err)
{
GtCondenserSearchArguments *arguments = tool_arguments;
int i, had_err = 0;
char *querypath = gt_str_get(arguments->querypath);
GtStr* coarse_fname = gt_str_new_cstr("coarse_");
char *db_basename = NULL;
char *suffix_ptr = NULL;
GtTimer *timer = NULL;
GtLogger *logger = NULL;
gt_error_check(err);
gt_assert(arguments);
logger = gt_logger_new(arguments->verbose, GT_LOGGER_DEFLT_PREFIX, stderr);
db_basename = gt_basename(gt_str_get(arguments->dbpath));
/* if first char is '.' this might be a hidden file */
if (strlen(db_basename) > (size_t) 1 &&
(suffix_ptr = strrchr(db_basename + 1, '.')) != NULL) {
/* remove suffix */
*suffix_ptr = '\0';
}
gt_str_append_cstr(coarse_fname, db_basename);
gt_str_append_cstr(coarse_fname, ".fas");
gt_free(db_basename);
db_basename = NULL;
suffix_ptr = NULL;
if (arguments->blastn || arguments->blastp) {
GtMatch *match;
GtMatchIterator *mp = NULL;
GtNREncseq *nrencseq = NULL;
GtStr *fastaname = gt_str_clone(arguments->dbpath);
HitPosition *hits;
double eval,
raw_eval = 0.0;
GtUword coarse_db_len = 0;
GtMatchIteratorStatus status;
int curr_hits = 0,
max_hits = 100;
hits = gt_malloc(sizeof (*hits) * (size_t) max_hits);
gt_str_append_cstr(fastaname, ".fas");
for (i=0; i < max_hits; i++) {
hits[i].range = gt_malloc(sizeof (*hits[i].range) * (size_t) 1);
}
if (gt_showtime_enabled()) {
timer = gt_timer_new_with_progress_description("initialization");
gt_timer_start(timer);
}
/*extract sequences from compressed database*/
if (!had_err) {
nrencseq = gt_n_r_encseq_new_from_file(gt_str_get(arguments->dbpath),
logger, err);
if (nrencseq == NULL)
had_err = -1;
}
if (!had_err) {
if (arguments->ceval == GT_UNDEF_DOUBLE ||
arguments->feval == GT_UNDEF_DOUBLE) {
/* from NCBI BLAST tutorial:
E = Kmne^{-lambdaS}
calculates E-value for score S with natural scale parameters K for
search space size and lambda for the scoring system
E = mn2^-S'
m being the subject (total) length, n the length of ONE query
calculates E-value for bit-score S'
*/
GtFastaReader *reader;
GtCondenserSearchAvg avg = {0,0};
reader = gt_fasta_reader_rec_new(arguments->querypath);
had_err = gt_fasta_reader_run(reader, NULL, NULL,
gt_condenser_search_cum_moving_avg,
&avg,
err);
if (!had_err) {
GtUword S = arguments->bitscore;
gt_log_log(GT_WU " queries, avg query size: " GT_WU,
avg.count, avg.avg);
raw_eval = 1/pow(2.0, (double) S) * avg.avg;
gt_logger_log(logger, "Raw E-value set to %.4e", raw_eval);
gt_assert(avg.avg != 0);
}
gt_fasta_reader_delete(reader);
}
}
/*create BLAST database from compressed database fasta file*/
if (!had_err) {
if (timer != NULL)
gt_timer_show_progress(timer, "create coarse BLAST db", stderr);
if (arguments->blastn)
had_err = gt_condenser_search_create_nucl_blastdb(gt_str_get(fastaname),
err);
else
had_err = gt_condenser_search_create_prot_blastdb(gt_str_get(fastaname),
err);
}
if (!had_err) {
GtBlastProcessCall *call;
if (timer != NULL)
gt_timer_show_progress(timer, "coarse BLAST run", stderr);
if (arguments->blastp)
call = gt_blast_process_call_new_prot();
else
call = gt_blast_process_call_new_nucl();
gt_blast_process_call_set_db(call, gt_str_get(fastaname));
gt_blast_process_call_set_query(call, querypath);
gt_blast_process_call_set_evalue(call, arguments->ceval);
gt_blast_process_call_set_num_threads(call, arguments->blthreads);
mp = gt_match_iterator_blast_process_new(call, err);
if (!mp)
had_err = -1;
gt_blast_process_call_delete(call);
while (!had_err &&
(status = gt_match_iterator_next(mp, &match, err)) !=
GT_MATCHER_STATUS_END)
{
if (status == GT_MATCHER_STATUS_OK) {
GtUword hit_seq_id;
char string[7];
const char *dbseqid = gt_match_get_seqid2(match);
if (sscanf(dbseqid,"%6s" GT_WU, string, &hit_seq_id) == 2) {
gt_match_get_range_seq2(match, hits[curr_hits].range);
hits[curr_hits].idx = hit_seq_id;
gt_match_delete(match);
curr_hits++;
if (curr_hits == max_hits) {
HitPosition *hit_extention;
max_hits += 100;
hits = gt_realloc(hits, sizeof (*hit_extention) * max_hits);
for (i=max_hits - 100; i < max_hits; i++) {
hits[i].range = gt_malloc(sizeof (*hits[i].range));
}
}
} else {
gt_error_set(err, "could not parse unique db header %s", dbseqid);
had_err = -1;
}
} else if (status == GT_MATCHER_STATUS_ERROR) {
had_err = -1;
}
}
gt_match_iterator_delete(mp);
}
/*extract sequences*/
if (!had_err) {
GtNREncseqDecompressor *decomp;
GtFile *coarse_hits;
if (timer != NULL)
gt_timer_show_progress(timer, "extract coarse search hits", stderr);
decomp = gt_n_r_encseq_decompressor_new(nrencseq);
coarse_hits = gt_file_new(gt_str_get(coarse_fname),"w", err);
/* TODO DW do NOT extract complete uniques! these could be complete
chromosomes!! just extract something around it? maybe +- max query
length*/
for (i = 0; i < curr_hits; i++) {
gt_n_r_encseq_decompressor_add_unique_idx_to_extract(decomp,
hits[i].idx);
}
had_err =
gt_n_r_encseq_decompressor_start_unique_extraction(coarse_hits,
decomp,
&coarse_db_len,
err);
gt_assert(coarse_db_len != 0);
gt_file_delete(coarse_hits);
gt_n_r_encseq_decompressor_delete(decomp);
}
gt_n_r_encseq_delete(nrencseq);
/* create BLAST database from decompressed database file */
if (!had_err) {
if (timer != NULL)
gt_timer_show_progress(timer, "create fine BLAST db", stderr);
if (arguments->blastn)
had_err =
gt_condenser_search_create_nucl_blastdb(gt_str_get(coarse_fname),
err);
else
had_err =
gt_condenser_search_create_prot_blastdb(gt_str_get(coarse_fname),
err);
}
/* perform fine BLAST search */
if (!had_err) {
GtBlastProcessCall *call;
if (timer != NULL)
gt_timer_show_progress(timer, "fine BLAST run", stderr);
if (arguments->feval == GT_UNDEF_DOUBLE) {
eval = raw_eval * coarse_db_len;
} else {
eval = arguments->feval;
}
if (arguments->blastp)
call = gt_blast_process_call_new_prot();
else
call = gt_blast_process_call_new_nucl();
gt_blast_process_call_set_db(call, gt_str_get(coarse_fname));
gt_blast_process_call_set_query(call, querypath);
gt_blast_process_call_set_evalue(call, eval);
gt_blast_process_call_set_num_threads(call, arguments->blthreads);
gt_logger_log(logger, "Fine E-value set to: %.4e (len)" GT_WU, eval,
coarse_db_len);
mp = gt_match_iterator_blast_process_new(call, err);
if (!mp)
had_err = -1;
gt_blast_process_call_delete(call);
if (!had_err) {
GtUword numofhits = 0;
while (!had_err &&
(status = gt_match_iterator_next(mp, &match, err)) !=
GT_MATCHER_STATUS_END) {
if (status == GT_MATCHER_STATUS_OK) {
GtMatchBlast *matchb = (GtMatchBlast*) match;
char *dbseqid = gt_malloc(sizeof (*dbseqid) * 50);
GtRange range_seq1;
GtRange range_seq2;
numofhits++;
gt_match_get_range_seq1(match, &range_seq1);
gt_match_get_range_seq2(match, &range_seq2);
gt_file_xprintf(
arguments->outfp,
"%s\t%s\t%.2f\t" GT_WU "\t" GT_WU "\t" GT_WU "\t" GT_WU "\t"
GT_WU "\t%g\t%.3f\n",
gt_match_get_seqid1(match),
gt_match_get_seqid2(match),
gt_match_blast_get_similarity(matchb),
gt_match_blast_get_align_length(matchb),
range_seq1.start,
range_seq1.end,
range_seq2.start,
range_seq2.end,
gt_match_blast_get_evalue(matchb),
(double) gt_match_blast_get_bitscore(matchb));
gt_match_delete(match);
gt_free(dbseqid);
} else if (status == GT_MATCHER_STATUS_ERROR) {
had_err = -1;
}
}
gt_log_log(GT_WU " hits found\n", numofhits);
}
gt_match_iterator_delete(mp);
}
if (!had_err)
if (timer != NULL)
gt_timer_show_progress_final(timer, stderr);
gt_timer_delete(timer);
/*cleanup*/
for (i=0; i < max_hits; i++) {
gt_free(hits[i].range);
}
gt_free(hits);
gt_str_delete(fastaname);
}
gt_str_delete(coarse_fname);
gt_logger_delete(logger);
return had_err;
}
static int cluster_sequences(GtArray *matches,
GtClusteredSet *cs,
GtHashmap *seqdesc2seqnum,
unsigned int psmall,
unsigned int plarge,
GtEncseq *encseq,
GtError *err)
{
GtMatch *match;
GtMatchEdgeTable matchedgetab;
GtMatchEdge matchedge;
GtRange rng_seq1,
rng_seq2;
int had_err = 0;
unsigned long i,
lsmall,
llarge,
matchlen1,
matchlen2,
num_of_seq,
seqnum1 = 0,
seqnum2 = 0;
const char *seqid;
num_of_seq = gt_encseq_num_of_sequences(encseq);
gt_assert(matches && cs && seqdesc2seqnum && encseq);
if (gt_clustered_set_num_of_elements(cs, err) != num_of_seq) {
had_err = -1;
gt_error_set(err,
"number of sequences (%lu) unequals number of elements in"
" clustered set (%lu)",
num_of_seq, gt_clustered_set_num_of_elements(cs, err));
}
if (!had_err) {
matchedgetab.edges = gt_array_new(sizeof (GtMatchEdge));
matchedgetab.num_of_edges = 0;
for (i = 0; i < gt_array_size(matches); i++) {
match = *(GtMatch**) gt_array_get(matches, i);
gt_match_get_range_seq1(match, &rng_seq1);
gt_match_get_range_seq2(match, &rng_seq2);
matchlen1 = gt_range_length(&rng_seq1);
matchlen2 = gt_range_length(&rng_seq2);
seqid = gt_match_get_seqid1(match);
if (gt_hashmap_get(seqdesc2seqnum, (void*) seqid) != NULL)
seqnum1 = ((unsigned long) gt_hashmap_get(seqdesc2seqnum, seqid)) - 1;
else {
had_err = -1;
gt_error_set(err, "key %s not found", seqid);
}
seqid = gt_match_get_seqid2(match);
if (!had_err && gt_hashmap_get(seqdesc2seqnum, (void*) seqid) != NULL)
seqnum2 = ((unsigned long) gt_hashmap_get(seqdesc2seqnum, seqid)) - 1;
else {
had_err = -1;
gt_error_set(err, "key %s not found", seqid);
}
if (!had_err) {
if (gt_encseq_seqlength(encseq, seqnum1) >
gt_encseq_seqlength(encseq, seqnum2)) {
llarge = gt_encseq_seqlength(encseq, seqnum1);
lsmall = gt_encseq_seqlength(encseq, seqnum2);
} else {
lsmall = gt_encseq_seqlength(encseq, seqnum1);
llarge = gt_encseq_seqlength(encseq, seqnum2);
}
if (((llarge * plarge)/100 <= matchlen1) &&
((lsmall * psmall)/100 <= matchlen1) &&
((llarge * plarge)/100 <= matchlen2) &&
((lsmall * psmall)/100 <= matchlen2)) {
if (seqnum1 != seqnum2) {
matchedge.matchnum0 = seqnum1;
matchedge.matchnum1 = seqnum2;
gt_array_add(matchedgetab.edges, matchedge);
matchedgetab.num_of_edges++;
}
}
}
}
}
if (!had_err)
if (gt_cluster_matches(cs, &matchedgetab, err) != 0)
had_err = -1;
if (!had_err)
gt_array_delete(matchedgetab.edges);
return had_err;
}
