static int gt_readjoiner_assembly_build_contained_reads_list(
GtReadjoinerAssemblyArguments *arguments, GtBitsequence **contained,
GtError *err)
{
int had_err = 0;
unsigned int i;
GtUword nofreads, nofreads_i;
GtStr *filename;
filename = gt_str_clone(arguments->readset);
gt_str_append_cstr(filename, ".0" GT_READJOINER_SUFFIX_CNTLIST);
had_err = gt_cntlist_parse(gt_str_get(filename), true, contained,
&nofreads, err);
for (i = 1U; i < arguments->nspmfiles && had_err == 0; i++)
{
gt_str_reset(filename);
gt_str_append_str(filename, arguments->readset);
gt_str_append_char(filename, '.');
gt_str_append_uint(filename, i);
gt_str_append_cstr(filename, GT_READJOINER_SUFFIX_CNTLIST);
had_err = gt_cntlist_parse(gt_str_get(filename), false, contained,
&nofreads_i, err);
gt_assert(had_err || nofreads == nofreads_i);
}
gt_str_delete(filename);
return had_err;
}
static int gt_readjoiner_assembly_count_spm(const char *readset, bool eqlen,
unsigned int minmatchlength, unsigned int nspmfiles, GtStrgraph *strgraph,
GtBitsequence *contained, GtLogger *default_logger, GtError *err)
{
GtSpmprocSkipData skipdata;
int had_err = 0;
unsigned int i;
GtStr *filename = gt_str_new();
gt_logger_log(default_logger, GT_READJOINER_ASSEMBLY_MSG_COUNTSPM);
if (!eqlen)
{
skipdata.out.e.proc = gt_spmproc_strgraph_count;
skipdata.to_skip = contained;
skipdata.out.e.data = strgraph;
}
for (i = 0; i < nspmfiles; i++)
{
gt_str_append_cstr(filename, readset);
gt_str_append_char(filename, '.');
gt_str_append_uint(filename, i);
gt_str_append_cstr(filename, GT_READJOINER_SUFFIX_SPMLIST);
had_err = gt_spmlist_parse(gt_str_get(filename),
(GtUword)minmatchlength,
eqlen ? gt_spmproc_strgraph_count : gt_spmproc_skip,
eqlen ? (void*)strgraph : (void*)&skipdata, err);
gt_str_reset(filename);
}
gt_str_delete(filename);
return had_err;
}
static void gff3_version_string(GtNodeVisitor *nv)
{
GtGFF3Visitor *gff3_visitor = gff3_visitor_cast(nv);
gt_assert(gff3_visitor);
if (!gff3_visitor->version_string_shown) {
if (!gff3_visitor->outstr) {
gt_file_xprintf(gff3_visitor->outfp, "%s %u\n", GT_GFF_VERSION_PREFIX,
GT_GFF_VERSION);
} else {
gt_str_append_cstr(gff3_visitor->outstr, GT_GFF_VERSION_PREFIX);
gt_str_append_cstr(gff3_visitor->outstr, " ");
gt_str_append_uint(gff3_visitor->outstr, GT_GFF_VERSION);
gt_str_append_char(gff3_visitor->outstr, '\n');
}
gff3_visitor->version_string_shown = true;
}
}
/* 'static' function */
GtStrArray* gt_trans_table_get_scheme_descriptions()
{
GtUword i;
GtTranslationScheme *scheme;
GtStr *str;
GtStrArray *sa = gt_str_array_new();
str = gt_str_new();
for (i = 1UL; i < (GtUword) GT_SIZEOFTRANSRANGE; i++) {
if (transnum2index[i] == GT_UNDEFTRANSNUM)
continue;
scheme = schemetable + transnum2index[i];
gt_str_reset(str);
gt_str_append_uint(str, scheme->identity);
gt_str_append_cstr(str, ": ");
gt_str_append_cstr(str, scheme->name);
gt_str_array_add_cstr(sa, gt_str_get(str));
}
gt_str_delete(str);
return sa;
}
GtNodeVisitor* gt_ltrdigest_pdom_visitor_new(GtPdomModelSet *model,
double eval_cutoff,
unsigned int chain_max_gap_length,
GtPdomCutoff cutoff,
GtRegionMapping *rmap,
GtError *err)
{
GtNodeVisitor *nv;
GtLTRdigestPdomVisitor *lv;
GtStr *cmd;
int had_err = 0, i, rval;
gt_assert(model && rmap);
rval = system("hmmscan -h > /dev/null");
if (rval == -1) {
gt_error_set(err, "error executing system(hmmscan)");
return NULL;
}
#ifndef _WIN32
if (WEXITSTATUS(rval) != 0) {
gt_error_set(err, "cannot find the hmmscan executable in PATH");
return NULL;
}
#else
/* XXX */
gt_error_set(err, "hmmscan for Windows not implemented");
return NULL;
#endif
nv = gt_node_visitor_create(gt_ltrdigest_pdom_visitor_class());
lv = gt_ltrdigest_pdom_visitor_cast(nv);
lv->eval_cutoff = eval_cutoff;
lv->cutoff = cutoff;
lv->chain_max_gap_length = chain_max_gap_length;
lv->rmap = rmap;
lv->output_all_chains = false;
lv->tag = gt_str_new_cstr("GenomeTools");
lv->root_type = gt_symbol(gt_ft_LTR_retrotransposon);
for (i = 0; i < 3; i++) {
lv->fwd[i] = gt_str_new();
lv->rev[i] = gt_str_new();
}
if (!had_err) {
cmd = gt_str_new_cstr("hmmscan --cpu ");
gt_str_append_uint(cmd, gt_jobs);
gt_str_append_cstr(cmd, " ");
switch (cutoff) {
case GT_PHMM_CUTOFF_GA:
gt_str_append_cstr(cmd, "--cut_ga");
break;
case GT_PHMM_CUTOFF_TC:
gt_str_append_cstr(cmd, "--cut_tc");
break;
case GT_PHMM_CUTOFF_NONE:
gt_str_append_cstr(cmd, "--domE ");
gt_str_append_double(cmd, eval_cutoff, 50);
break;
}
gt_str_append_cstr(cmd, " ");
gt_str_append_cstr(cmd, gt_pdom_model_set_get_filename(model));
gt_str_append_cstr(cmd, " -");
lv->cmdline = cmd;
lv->args = gt_cstr_split(gt_str_get(lv->cmdline), ' ');
gt_log_log("HMMER cmdline: %s", gt_str_get(cmd));
}
return nv;
}
