static int gt_ltrdigest_pdom_visitor_parse_alignments(GT_UNUSED
GtLTRdigestPdomVisitor
*lv,
GtHMMERParseStatus *status,
char *buf,
FILE *instream,
GtError *err)
{
int had_err = 0, cur_domain = GT_UNDEF_INT, line = -1;
GtHMMERSingleHit *hit = NULL;
gt_assert(lv && instream && status);
gt_error_check(err);
had_err = pdom_parser_get_next_line(buf, instream, err);
gt_assert(buf != NULL);
while (!had_err && strncmp("Internal pipeline statistics",
buf, (size_t) 28) &&
strncmp(">>", buf, (size_t) 2)) {
if ((buf[2] == '=' && buf[3] == '=')) {
buf[17] = '\0';
cur_domain = atoi(buf+12);
gt_assert(cur_domain != GT_UNDEF_INT && cur_domain > 0);
hit = gt_hmmer_parse_status_get_hit(status,
(unsigned long) cur_domain - 1);
gt_assert(hit && !hit->alignment);
hit->alignment = gt_str_new();
hit->aastring = gt_str_new();
line = -2;
} else {
gt_assert(hit && hit->alignment);
gt_str_append_cstr(hit->alignment, buf);
gt_str_append_char(hit->alignment, '\n');
switch (line % 4) {
case 1:
gt_str_append_char(hit->alignment, '\n');
break;
case 0:
{
char *b = buf;
b = strtok(buf, " ");
gt_assert(strspn(b, "012+-") == (size_t) 2);
b = strtok(NULL, " ");
gt_assert(strlen(b) > 0);
b = strtok(NULL, " ");
gt_ltrdigest_pdom_visitor_add_aaseq(b, hit->aastring);
}
break;
}
line++;
}
had_err = pdom_parser_get_next_line(buf, instream, err);
}
return had_err;
}
static int gt_ltrdigest_pdom_visitor_parse_query(GtLTRdigestPdomVisitor *lv,
GtHMMERParseStatus *status,
bool *end,
FILE *instream, GtError *err)
{
int had_err = 0;
char buf[GT_HMMER_BUF_LEN];
gt_assert(lv && instream && status);
gt_error_check(err);
had_err = pdom_parser_get_next_line(buf, instream, err);
if (!had_err && strncmp("Query:", buf, (size_t) 6) != 0) {
*end = true;
}
if (!had_err && !(*end)) {
status->strand = gt_strand_get(buf[14]);
buf[14] = '\0';
status->frame = (unsigned) atoi(buf+13);
}
if (!had_err && !(*end)) {
had_err = gt_ltrdigest_pdom_visitor_parse_scores(lv, buf, instream, err);
}
if (!had_err && !(*end)) {
had_err = gt_ltrdigest_pdom_visitor_parse_domainhits(lv, status, buf,
instream, err);
}
if (!had_err && !(*end)) {
had_err = gt_ltrdigest_pdom_visitor_parse_statistics(lv, buf, instream,
err);
}
return had_err;
}
static int gt_ltrdigest_pdom_visitor_parse_scores(GT_UNUSED
GtLTRdigestPdomVisitor *lv,
char *buf, FILE *instream,
GtError *err)
{
int had_err = 0;
gt_assert(lv && instream);
gt_error_check(err);
had_err = pdom_parser_get_next_line(buf, instream, err);
if (!had_err && strncmp("Scores", buf, (size_t) 6) != 0) {
gt_error_set(err, "expected 'Scores:' at beginning of new scores "
"section, '%s' read instead", buf);
had_err = -1;
}
while (!had_err && strncmp("Domain annotation", buf, (size_t) 17))
had_err = pdom_parser_get_next_line(buf, instream, err);
return had_err;
}
static int gt_ltrdigest_pdom_visitor_parse_statistics(GT_UNUSED
GtLTRdigestPdomVisitor
*lv,
char *buf, FILE *instream,
GtError *err)
{
int had_err = 0;
gt_assert(lv && instream);
gt_error_check(err);
while (!had_err && (buf[0] != '/' && buf[1] != '/'))
had_err = pdom_parser_get_next_line(buf, instream, err);
return had_err;
}
static int gt_ltrdigest_pdom_visitor_parse_domainhits(GtLTRdigestPdomVisitor
*lv,
GtHMMERParseStatus *status,
char *buf,
FILE *instream,
GtError *err)
{
int had_err = 0;
GtUword i, nof_targets = 0, nof_hits = 0;
gt_assert(lv && instream && status);
gt_error_check(err);
had_err = pdom_parser_get_next_line(buf, instream, err);
gt_assert(buf != NULL);
while (!had_err && strncmp("Internal", buf, (size_t) 8)) {
GtUword no, hmmfrom, hmmto, alifrom, alito;
double score, evalue;
char threshold_ok = '-';
if ((buf[0] == '>' && buf[1] == '>')) {
char *b = buf;
b = strtok(buf+3, " ");
gt_str_reset(status->cur_model);
gt_str_append_cstr(status->cur_model, b);
had_err = pdom_parser_get_next_line(buf, instream, err);
if (!had_err && strncmp(" [No individual", buf, (size_t) 17)) {
for (i = 0UL; i < 2UL && !had_err; i++)
had_err = pdom_parser_get_next_line(buf, instream, err);
}
nof_targets++;
nof_hits = 0UL;
gt_hmmer_parse_status_mark_frame_finished(status);
}
while (!had_err &&
8 == sscanf(buf, ""GT_WU" %c %lf %*f %*f %lf "GT_WU" "GT_WU" %*s "
GT_WU" "GT_WU"", &no, &threshold_ok, &score, &evalue,
&hmmfrom, &hmmto, &alifrom, &alito)) {
GtHMMERSingleHit *shit = gt_calloc((size_t) 1, sizeof (*shit));
shit->hmmfrom = hmmfrom;
shit->hmmto = hmmto;
shit->alifrom = alifrom;
shit->alito = alito;
shit->score = score;
shit->evalue = evalue;
shit->strand = status->strand;
shit->frame = (GtUword) status->frame;
shit->reported = (threshold_ok == '!');
shit->chains = gt_array_new(sizeof (GtUword));
gt_hmmer_parse_status_add_hit(status, shit);
nof_hits++;
had_err = pdom_parser_get_next_line(buf, instream, err);
}
if (!had_err) {
if (nof_hits > 0)
had_err = gt_ltrdigest_pdom_visitor_parse_alignments(lv, status, buf,
instream, err);
else
had_err = pdom_parser_get_next_line(buf, instream, err);
}
}
return had_err;
}
static int gt_ltrdigest_pdom_visitor_parse_alignments(GT_UNUSED
GtLTRdigestPdomVisitor
*lv,
GtHMMERParseStatus *status,
char *buf,
FILE *instream,
GtError *err)
{
int had_err = 0, cur_domain = GT_UNDEF_INT, line = GT_UNDEF_INT;
bool first_align_line = false;
int mod_val = 4;
GtHMMERSingleHit *hit = NULL;
gt_assert(lv && instream && status);
gt_error_check(err);
had_err = pdom_parser_get_next_line(buf, instream, err);
gt_assert(buf != NULL);
while (!had_err && strncmp("Internal pipeline statistics",
buf, (size_t) 28) &&
strncmp(">>", buf, (size_t) 2)) {
if ((buf[2] == '=' && buf[3] == '=')) {
buf[17] = '\0';
cur_domain = atoi(buf+12);
gt_assert(cur_domain != GT_UNDEF_INT && cur_domain > 0);
hit = gt_hmmer_parse_status_get_hit(status,
(GtUword) cur_domain - 1);
gt_assert(hit && !hit->alignment);
hit->alignment = gt_str_new();
hit->aastring = gt_str_new();
first_align_line = true;
mod_val = 4;
} else {
bool run = true;
char junkbuf[BUFSIZ];
if (first_align_line) {
/* some models contain consensus structure annotation -- in this case
there is an additional line in the output which must be taken
into account */
line = 0;
if (1 == sscanf(buf, "%*s %s", junkbuf)) {
if (0 == strcmp(junkbuf, "CS") || 0 == strcmp(junkbuf, "RF")) {
mod_val = 5;
line = -1;
run = false;
}
}
first_align_line = false;
}
if (run) {
gt_assert(hit && hit->alignment);
gt_str_append_cstr(hit->alignment, buf);
gt_str_append_char(hit->alignment, '\n');
switch (line % mod_val) {
case 1:
gt_str_append_char(hit->alignment, '\n');
break;
case 2:
{
GT_UNUSED char *b = buf;
b = strtok(buf, " ");
gt_assert(strspn(b, "012+-") == (size_t) 2);
b = strtok(NULL, " ");
gt_assert(strlen(b) > 0);
b = strtok(NULL, " ");
gt_ltrdigest_pdom_visitor_add_aaseq(b, hit->aastring);
}
break;
}
}
line++;
}
had_err = pdom_parser_get_next_line(buf, instream, err);
}
return had_err;
}