static GtArray*
gaeval_visitor_intersect(GtGenomeNode *genemodel, GtGenomeNode *alignment)
{
agn_assert(genemodel && alignment);
GtFeatureNode *genefn = gt_feature_node_cast(genemodel);
GtFeatureNode *algnfn = gt_feature_node_cast(alignment);
agn_assert(gt_feature_node_has_type(genefn, "mRNA"));
GtStrand genestrand = gt_feature_node_get_strand(genefn);
GtStrand algnstrand = gt_feature_node_get_strand(algnfn);
if(genestrand != algnstrand)
return NULL;
GtArray *covered_parts = gt_array_new( sizeof(GtRange) );
GtArray *exons = agn_typecheck_select(genefn, agn_typecheck_exon);
GtWord i;
for(i = 0; i < gt_array_size(exons); i++)
{
GtGenomeNode *exon = *(GtGenomeNode **)gt_array_get(exons, i);
GtRange exonrange = gt_genome_node_get_range(exon);
GtFeatureNodeIterator *aniter = gt_feature_node_iterator_new(algnfn);
GtFeatureNode *tempaln;
GtRange nullrange = {0, 0};
for(tempaln = gt_feature_node_iterator_next(aniter);
tempaln != NULL;
tempaln = gt_feature_node_iterator_next(aniter))
{
if(gt_feature_node_has_type(tempaln, "match_gap"))
continue;
GtRange alnrange = gt_genome_node_get_range((GtGenomeNode *) tempaln);
GtRange intr = gaeval_visitor_range_intersect(&exonrange, &alnrange);
if(gt_range_compare(&intr, &nullrange) != 0)
gt_array_add(covered_parts, intr);
}
gt_feature_node_iterator_delete(aniter);
}
gt_array_delete(exons);
for(i = 0; i < gt_array_size(covered_parts); i++)
{
GtRange *r1 = gt_array_get(covered_parts, i);
GtUword j;
for(j = i+1; j < gt_array_size(covered_parts); j++)
{
GtRange *r2 = gt_array_get(covered_parts, j);
agn_assert(gt_range_overlap(r1, r2) == false);
}
}
return covered_parts;
}
static void feature_node_change_seqid(GtGenomeNode *gn, GtStr *seqid)
{
GtFeatureNode *fn = gt_feature_node_cast(gn);
gt_assert(fn && seqid);
gt_str_delete(fn->seqid);
fn->seqid = gt_str_ref(seqid);
}
GtGenomeNode* gt_feature_node_new(GtStr *seqid, const char *type,
GtUword start, GtUword end,
GtStrand strand)
{
GtGenomeNode *gn;
GtFeatureNode *fn;
gt_assert(seqid && type);
gt_assert(start <= end);
gn = gt_genome_node_create(gt_feature_node_class());
fn = gt_feature_node_cast(gn);
fn->seqid = gt_str_ref(seqid);
fn->source = NULL;
fn->type = gt_symbol(type);
fn->score = GT_UNDEF_FLOAT;
fn->range.start = start;
fn->range.end = end;
fn->representative = NULL;
fn->attributes = NULL;
fn->bit_field = 0;
fn->bit_field |= strand << STRAND_OFFSET;
fn->children = NULL; /* the children list is create on demand */
fn->observer = NULL;
gt_feature_node_set_phase(fn, GT_PHASE_UNDEFINED);
set_transcriptfeaturetype(fn, TRANSCRIPT_FEATURE_TYPE_UNDETERMINED);
set_tree_status(&fn->bit_field, IS_TREE);
/* the DFS status is set to DFS_WHITE already */
fn->representative = NULL;
return gn;
}
static int feature_index_lua_add_feature_node(lua_State *L)
{
GtFeatureIndex **fi;
GtGenomeNode **gn;
GtFeatureNode *fn;
GtStr *seqid;
GtError *err;
bool has_seqid;
gt_assert(L);
fi = check_feature_index(L, 1);
gn = check_genome_node(L, 2);
fn = gt_feature_node_cast(*gn);
luaL_argcheck(L, fn, 2, "not a feature node");
seqid = gt_genome_node_get_seqid(*gn);
luaL_argcheck(L, seqid, 2, "feature does not have a sequence id");
err = gt_error_new();
if (gt_feature_index_has_seqid(*fi, &has_seqid, gt_str_get(seqid), err))
return gt_lua_error(L, err);
gt_error_delete(err);
luaL_argcheck(L, has_seqid, 2,
"feature index does not contain corresponding sequence region");
err = gt_error_new();
if (gt_feature_index_add_feature_node(*fi, fn, err))
return gt_lua_error(L, err);
gt_error_delete(err);
return 0;
}
int gt_diagram_unit_test(GtError *err)
{
int had_err = 0;
GtGenomeNode *gn;
GtDiagramTestShared sh;
GtRange testrng = {100, 10000};
gt_error_check(err);
gn = gt_feature_node_new_standard_gene();
sh.fi = gt_feature_index_memory_new();
sh.sty = gt_style_new(err);
sh.err = err;
sh.errstatus = 0;
gt_feature_index_add_feature_node(sh.fi, gt_feature_node_cast(gn), err);
gt_genome_node_delete(gn);
sh.d = gt_diagram_new(sh.fi, "ctg123", &testrng, sh.sty, err);
/* removed the multithreading test for now until it is fixed */
gt_diagram_unit_test_sketch_func(&sh);
gt_ensure(sh.errstatus == 0);
gt_style_delete(sh.sty);
gt_diagram_delete(sh.d);
gt_feature_index_delete(sh.fi);
return had_err;
}
static void feature_node_set_range(GtGenomeNode *gn, const GtRange *range)
{
GtFeatureNode *fn = gt_feature_node_cast(gn);
fn->range = *range;
if (fn->observer && fn->observer->range_changed)
fn->observer->range_changed(fn, &(fn->range), fn->observer->data);
}
static GtFeatureIndex *in_stream_test_data(GtError *error)
{
GtFeatureIndex *fi;
GtFeatureNode *fn;
GtGenomeNode *gn;
GtRegionNode *rn;
GtStr *seqid;
fi = gt_feature_index_memory_new();
seqid = gt_str_new_cstr("chr1");
gn = gt_region_node_new(seqid, 1, 100000);
rn = gt_region_node_cast(gn);
gt_feature_index_add_region_node(fi, rn, error);
gt_genome_node_delete(gn);
gn = gt_feature_node_new(seqid, "region", 500, 5000, GT_STRAND_BOTH);
fn = gt_feature_node_cast(gn);
gt_feature_index_add_feature_node(fi, fn, error);
gt_genome_node_delete(gn);
gn = gt_feature_node_new(seqid, "region", 50000, 75000, GT_STRAND_BOTH);
fn = gt_feature_node_cast(gn);
gt_feature_index_add_feature_node(fi, fn, error);
gt_genome_node_delete(gn);
gt_str_delete(seqid);
seqid = gt_str_new_cstr("scf0001");
gn = gt_region_node_new(seqid, 1, 10000);
rn = gt_region_node_cast(gn);
gt_feature_index_add_region_node(fi, rn, error);
gt_genome_node_delete(gn);
gn = gt_feature_node_new(seqid, "mRNA", 4000, 6000, GT_STRAND_REVERSE);
fn = gt_feature_node_cast(gn);
gt_feature_index_add_feature_node(fi, fn, error);
gt_genome_node_delete(gn);
gn = gt_feature_node_new(seqid, "mRNA", 7000, 9500, GT_STRAND_FORWARD);
fn = gt_feature_node_cast(gn);
gt_feature_index_add_feature_node(fi, fn, error);
gt_genome_node_delete(gn);
gt_str_delete(seqid);
return fi;
}
static int feature_node_accept(GtGenomeNode *gn, GtNodeVisitor *nv,
GtError *err)
{
GtFeatureNode *fn;
gt_error_check(err);
fn = gt_feature_node_cast(gn);
return gt_node_visitor_visit_feature_node(nv, fn, err);
}
static int genome_node_lua_contains_marked(lua_State *L)
{
GtGenomeNode **gn;
gn = check_genome_node(L, 1);
lua_pushboolean(L,
gt_feature_node_contains_marked(gt_feature_node_cast(*gn)));
return 1;
}
GtGenomeNode* gt_feature_node_new_pseudo(GtStr *seqid, GtUword start,
GtUword end, GtStrand strand)
{
GtFeatureNode *pf;
GtGenomeNode *pn;
gt_assert(seqid);
gt_assert(start <= end);
pn = gt_feature_node_new(seqid, "pseudo", start, end, strand);
pf = gt_feature_node_cast(pn);
pf->type = NULL; /* pseudo features do not have a type */
pf->bit_field |= 1 << PSEUDO_FEATURE_OFFSET;
return pn;
}
GtGenomeNode* gt_feature_node_new_pseudo_template(GtFeatureNode *fn)
{
GtFeatureNode *pf;
GtGenomeNode *pn;
GtRange range;
gt_assert(fn);
range = feature_node_get_range((GtGenomeNode*) fn),
pn = gt_feature_node_new_pseudo(feature_node_get_seqid((GtGenomeNode*) fn),
range.start, range.end,
gt_feature_node_get_strand(fn));
pf = gt_feature_node_cast(pn);
gt_feature_node_set_source(pf, fn->source);
return pn;
}
static int extracttarget_from_node(GtGenomeNode *gn, GtStrArray *seqfiles,
GtError *err)
{
GtFeatureNodeIterator *fni;
int had_err = 0;
gt_error_check(err);
gt_assert(gn && seqfiles);
if (gt_genome_node_cast(gt_feature_node_class(), gn)) {
const char *target;
GtFeatureNode *child;
fni = gt_feature_node_iterator_new(gt_feature_node_cast(gn));
while (!had_err && /* XXX remove cast */
(child = (GtFeatureNode*) gt_feature_node_iterator_next(fni))) {
if ((target = gt_feature_node_get_attribute(child, "Target")))
had_err = extracttarget_from_seqfiles(target, seqfiles, err);
}
gt_feature_node_iterator_delete(fni);
}
return had_err;
}
static void feature_node_free(GtGenomeNode *gn)
{
GtFeatureNode *fn = gt_feature_node_cast(gn);
gt_str_delete(fn->seqid);
gt_str_delete(fn->source);
gt_tag_value_map_delete(fn->attributes);
if (fn->children) {
GtDlistelem *dlistelem;
for (dlistelem = gt_dlist_first(fn->children);
dlistelem != NULL;
dlistelem = gt_dlistelem_next(dlistelem)) {
gt_genome_node_delete(gt_dlistelem_get_data(dlistelem));
}
}
gt_dlist_delete(fn->children);
if (fn->observer && fn->observer->deleted)
fn->observer->deleted(fn, fn->observer->data);
if (fn->observer)
gt_feature_node_observer_delete(fn->observer);
}
static int extract_join_feature(GtGenomeNode *gn, const char *type,
GtRegionMapping *region_mapping,
GtStr *sequence, bool *reverse_strand,
bool *first_child_of_type_seen, GtPhase *phase,
GtError *err)
{
char *outsequence;
GtFeatureNode *fn;
GtRange range;
int had_err = 0;
gt_error_check(err);
fn = gt_feature_node_cast(gn);
gt_assert(fn);
if (gt_feature_node_has_type(fn, type)) {
if (gt_feature_node_get_strand(fn) == GT_STRAND_REVERSE) {
*reverse_strand = true;
*phase = gt_feature_node_get_phase(fn);
} else {
if (!(*first_child_of_type_seen)) {
*first_child_of_type_seen = true;
*phase = gt_feature_node_get_phase(fn);
} else *phase = GT_PHASE_UNDEFINED;
}
range = gt_genome_node_get_range(gn);
had_err = gt_region_mapping_get_sequence(region_mapping, &outsequence,
gt_genome_node_get_seqid(gn),
range.start, range.end, err);
if (!had_err) {
gt_str_append_cstr_nt(sequence, outsequence, gt_range_length(&range));
gt_free(outsequence);
}
}
return had_err;
}
/* to be called from implementing class! */
int gt_feature_index_unit_test(GtFeatureIndex *fi, GtError *err)
{
int had_err = 0, i, rval;
GtFeatureIndexTestShared sh;
GtStrArray *seqids;
GtStr *seqid;
GtRange check_range;
GtRegionNode *rn;
bool has_seqid;
gt_error_check(err);
sh.mutex = gt_mutex_new();
sh.nodes = gt_array_new(sizeof (GtFeatureNode*));
sh.error_count = 0;
sh.next_node_idx = 0;
sh.fi = fi;
sh.err = gt_error_new();
/* create region */
seqid = gt_str_new_cstr(GT_FI_TEST_SEQID);
rn = (GtRegionNode*) gt_region_node_new(seqid, GT_FI_TEST_START,
GT_FI_TEST_END);
/* test seqid is not supposed to exist */
gt_ensure(gt_feature_index_has_seqid(sh.fi, &has_seqid,
GT_FI_TEST_SEQID, err) == 0);
gt_ensure(!has_seqid);
/* add a sequence region directly and check if it has been added */
rval = gt_feature_index_add_region_node(sh.fi, rn, err);
gt_ensure(rval == 0);
gt_genome_node_delete((GtGenomeNode*) rn);
gt_ensure(gt_feature_index_has_seqid(sh.fi, &has_seqid,
GT_FI_TEST_SEQID, err) == 0);
gt_ensure(has_seqid);
gt_feature_index_get_range_for_seqid(sh.fi, &check_range, GT_FI_TEST_SEQID,
err);
gt_ensure(check_range.start == GT_FI_TEST_START
&& check_range.end == GT_FI_TEST_END);
/* set up nodes to store */
for (i=0;i<GT_FI_TEST_FEATURES_PER_THREAD*gt_jobs;i++) {
GtUword start, end;
GtFeatureNode *fn;
start = random() % (GT_FI_TEST_END - GT_FI_TEST_FEATURE_WIDTH);
end = start + random() % (GT_FI_TEST_FEATURE_WIDTH);
fn = gt_feature_node_cast(gt_feature_node_new(seqid, "gene", start, end,
GT_STRAND_FORWARD));
gt_array_add(sh.nodes, fn);
}
/* test parallel addition */
gt_multithread(gt_feature_index_unit_test_add, &sh, err);
seqids = gt_feature_index_get_seqids(fi, err);
gt_ensure(seqids);
gt_ensure(gt_feature_index_has_seqid(fi, &has_seqid,GT_FI_TEST_SEQID,
err) == 0);
gt_ensure(has_seqid);
gt_ensure(gt_str_array_size(seqids) == 1);
/* test parallel query */
if (!had_err)
gt_multithread(gt_feature_index_unit_test_query, &sh, err);
gt_ensure(sh.error_count == 0);
gt_mutex_delete(sh.mutex);
gt_error_delete(sh.err);
gt_str_array_delete(seqids);
gt_array_delete(sh.nodes);
gt_str_delete(seqid);
return had_err;
}
static int genome_node_lua_mark(lua_State *L)
{
GtGenomeNode **gn = check_genome_node(L, 1);
gt_feature_node_mark(gt_feature_node_cast(*gn));
return 0;
}
static int gt_extract_feature_sequence_generic(GtStr *sequence,
GtGenomeNode *gn,
const char *type, bool join, GtStr *seqid,
GtStrArray *target_ids,
unsigned int *out_phase_offset,
GtRegionMapping *region_mapping, GtError *err)
{
GtFeatureNode *fn;
GtRange range;
unsigned int phase_offset = 0;
char *outsequence;
const char *target;
int had_err = 0;
gt_error_check(err);
fn = gt_genome_node_cast(gt_feature_node_class(), gn);
gt_assert(fn);
if (seqid)
gt_str_append_str(seqid, gt_genome_node_get_seqid(gn));
if (target_ids &&
(target = gt_feature_node_get_attribute(fn, GT_GFF_TARGET))) {
had_err = gt_gff3_parser_parse_all_target_attributes(target, false,
target_ids, NULL,
NULL, "", 0, err);
}
if (!had_err) {
if (join) {
GtFeatureNodeIterator *fni;
GtFeatureNode *child;
bool reverse_strand = false,
first_child = true,
first_child_of_type_seen = false;
GtPhase phase = GT_PHASE_UNDEFINED;
/* in this case we have to traverse the children */
fni = gt_feature_node_iterator_new_direct(gt_feature_node_cast(gn));
while (!had_err && (child = gt_feature_node_iterator_next(fni))) {
if (first_child) {
if (target_ids &&
(target = gt_feature_node_get_attribute(child, GT_GFF_TARGET))) {
gt_str_array_reset(target_ids);
had_err = gt_gff3_parser_parse_all_target_attributes(target, false,
target_ids,
NULL,
NULL, "", 0,
err);
}
first_child = false;
}
if (!had_err) {
if (extract_join_feature((GtGenomeNode*) child, type, region_mapping,
sequence, &reverse_strand,
&first_child_of_type_seen,
&phase, err)) {
had_err = -1;
}
if (phase != GT_PHASE_UNDEFINED) {
phase_offset = (int) phase;
}
}
}
gt_feature_node_iterator_delete(fni);
gt_assert(phase_offset <= (unsigned int) GT_PHASE_UNDEFINED);
if (!had_err && gt_str_length(sequence)) {
if (reverse_strand) {
had_err = gt_reverse_complement(gt_str_get(sequence),
gt_str_length(sequence), err);
}
}
}
else if (gt_feature_node_get_type(fn) == type) {
GtPhase phase = gt_feature_node_get_phase(fn);
gt_assert(!had_err);
if (phase != GT_PHASE_UNDEFINED)
phase_offset = (unsigned int) phase;
/* otherwise we only have to look at this feature */
range = gt_genome_node_get_range(gn);
gt_assert(range.start); /* 1-based coordinates */
had_err = gt_region_mapping_get_sequence(region_mapping, &outsequence,
gt_genome_node_get_seqid(gn),
range.start, range.end, err);
if (!had_err) {
gt_str_append_cstr_nt(sequence, outsequence, gt_range_length(&range));
gt_free(outsequence);
if (gt_feature_node_get_strand(fn) == GT_STRAND_REVERSE) {
had_err = gt_reverse_complement(gt_str_get(sequence),
gt_str_length(sequence), err);
}
}
}
}
if (out_phase_offset && phase_offset != GT_PHASE_UNDEFINED) {
*out_phase_offset = phase_offset;
}
return had_err;
}
static GtStr* feature_node_get_seqid(GtGenomeNode *gn)
{
GtFeatureNode *fn = gt_feature_node_cast(gn);
return fn->seqid;
}
static GtRange feature_node_get_range(GtGenomeNode *gn)
{
GtFeatureNode *fn = gt_feature_node_cast(gn);
return fn->range;
}