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;
}
static int feature_index_lua_get_features_for_range(lua_State *L)
{
GtFeatureIndex **feature_index;
const char *seqid;
GtRange *range;
GtError *err;
bool has_seqid;
GtArray *features;
GT_UNUSED int had_err;
feature_index = check_feature_index(L, 1);
seqid = luaL_checkstring(L, 2);
err = gt_error_new();
if (gt_feature_index_has_seqid(*feature_index, &has_seqid, seqid, err))
return gt_lua_error(L, err);
gt_error_delete(err);
luaL_argcheck(L, has_seqid, 2,
"feature_index does not contain seqid");
range = check_range(L, 3);
features = gt_array_new(sizeof (GtGenomeNode*));
err = gt_error_new();
had_err = gt_feature_index_get_features_for_range(*feature_index, features,
seqid, range, err);
if (had_err)
return gt_lua_error(L, err);
gt_error_delete(err);
push_features_as_table(L, features);
gt_array_delete(features);
return 1;
}
GtStrArray* agn_seq_union(GtFeatureIndex *refrfeats, GtFeatureIndex *predfeats,
AgnLogger *logger)
{
// Fetch seqids from reference and prediction annotations
GtError *e = gt_error_new();
GtStrArray *refrseqids = gt_feature_index_get_seqids(refrfeats, e);
if(gt_error_is_set(e))
{
agn_logger_log_error(logger, "error fetching seqids for reference: %s",
gt_error_get(e));
gt_error_unset(e);
}
GtStrArray *predseqids = gt_feature_index_get_seqids(predfeats, e);
if(gt_error_is_set(e))
{
agn_logger_log_error(logger, "error fetching seqids for prediction: %s",
gt_error_get(e));
gt_error_unset(e);
}
gt_error_delete(e);
if(agn_logger_has_error(logger))
{
gt_str_array_delete(refrseqids);
gt_str_array_delete(predseqids);
return NULL;
}
GtStrArray *seqids = agn_gt_str_array_union(refrseqids, predseqids);
gt_str_array_delete(refrseqids);
gt_str_array_delete(predseqids);
return seqids;
}
static void infer_cds_visitor_test_data(GtQueue *queue)
{
GtError *error = gt_error_new();
const char *file = "data/gff3/grape-codons.gff3";
GtNodeStream *gff3in = gt_gff3_in_stream_new_unsorted(1, &file);
gt_gff3_in_stream_check_id_attributes((GtGFF3InStream *)gff3in);
gt_gff3_in_stream_enable_tidy_mode((GtGFF3InStream *)gff3in);
GtLogger *logger = gt_logger_new(true, "", stderr);
GtNodeStream *icv_stream = agn_infer_cds_stream_new(gff3in, NULL, logger);
GtArray *feats = gt_array_new( sizeof(GtFeatureNode *) );
GtNodeStream *arraystream = gt_array_out_stream_new(icv_stream, feats, error);
int pullresult = gt_node_stream_pull(arraystream, error);
if(pullresult == -1)
{
fprintf(stderr, "[AgnInferCDSVisitor::infer_cds_visitor_test_data] error "
"processing features: %s\n", gt_error_get(error));
}
gt_node_stream_delete(gff3in);
gt_node_stream_delete(icv_stream);
gt_node_stream_delete(arraystream);
gt_logger_delete(logger);
gt_array_sort(feats, (GtCompare)agn_genome_node_compare);
gt_array_reverse(feats);
while(gt_array_size(feats) > 0)
{
GtFeatureNode *fn = *(GtFeatureNode **)gt_array_pop(feats);
gt_queue_add(queue, fn);
}
gt_array_delete(feats);
gt_error_delete(error);
}
GtFeatureIndex *agn_import_canonical(int numfiles, const char **filenames,
AgnLogger *logger)
{
GtNodeStream *gff3 = gt_gff3_in_stream_new_unsorted(numfiles, filenames);
gt_gff3_in_stream_check_id_attributes((GtGFF3InStream *)gff3);
gt_gff3_in_stream_enable_tidy_mode((GtGFF3InStream *)gff3);
GtFeatureIndex *features = gt_feature_index_memory_new();
GtNodeStream *cgstream = agn_canon_gene_stream_new(gff3, logger);
GtNodeStream *featstream = gt_feature_out_stream_new(cgstream, features);
GtError *error = gt_error_new();
int result = gt_node_stream_pull(featstream, error);
if(result == -1)
{
agn_logger_log_error(logger, "error processing node stream: %s",
gt_error_get(error));
}
gt_error_delete(error);
if(agn_logger_has_error(logger))
{
gt_feature_index_delete(features);
features = NULL;
}
gt_node_stream_delete(gff3);
gt_node_stream_delete(cgstream);
gt_node_stream_delete(featstream);
return features;
}
GtFeatureIndex *agn_import_simple(int numfiles, const char **filenames,
char *type, AgnLogger *logger)
{
GtFeatureIndex *features = gt_feature_index_memory_new();
GtNodeStream *gff3 = gt_gff3_in_stream_new_unsorted(numfiles, filenames);
gt_gff3_in_stream_check_id_attributes((GtGFF3InStream *)gff3);
gt_gff3_in_stream_enable_tidy_mode((GtGFF3InStream *)gff3);
GtHashmap *typestokeep = gt_hashmap_new(GT_HASH_STRING, NULL, NULL);
gt_hashmap_add(typestokeep, type, type);
GtNodeStream *filterstream = agn_filter_stream_new(gff3, typestokeep);
GtNodeStream *featstream = gt_feature_out_stream_new(filterstream, features);
GtError *error = gt_error_new();
int result = gt_node_stream_pull(featstream, error);
if(result == -1)
{
agn_logger_log_error(logger, "error processing node stream: %s",
gt_error_get(error));
}
gt_error_delete(error);
if(agn_logger_has_error(logger))
{
gt_feature_index_delete(features);
features = NULL;
}
gt_node_stream_delete(gff3);
gt_node_stream_delete(filterstream);
gt_node_stream_delete(featstream);
return features;
}
static int feature_node_lua_extract_sequence(lua_State *L)
{
GtGenomeNode **gn;
GtFeatureNode *fn;
const char *type;
bool join;
GtRegionMapping **region_mapping;
GtStr *sequence;
GtError *err;
gn = check_genome_node(L, 1);
/* make sure we get a feature node */
fn = gt_feature_node_try_cast(*gn);
luaL_argcheck(L, fn, 1, "not a feature node");
type = luaL_checkstring(L, 2);
join = lua_toboolean(L, 3);
region_mapping = check_region_mapping(L, 4);
err = gt_error_new();
sequence = gt_str_new();
if (gt_extract_feature_sequence(sequence, *gn, type, join, NULL, NULL,
*region_mapping, err)) {
gt_str_delete(sequence);
return gt_lua_error(L, err);
}
if (gt_str_length(sequence))
lua_pushstring(L, gt_str_get(sequence));
else
lua_pushnil(L);
gt_str_delete(sequence);
gt_error_delete(err);
return 1;
}
GtUword gt_cntlist_xload(const char *filename, GtBitsequence **cntlist,
GtUword expected_nofreads)
{
int retval;
GtUword found_nofreads;
GtError *err;
if (!gt_file_exists(filename))
{
fprintf(stderr, "FATAL: error by loading contained reads list: "
"file %s does not exist\n", filename);
exit(EXIT_FAILURE);
}
err = gt_error_new();
retval = gt_cntlist_parse(filename, true, cntlist, &found_nofreads, err);
if (retval != 0)
{
fprintf(stderr, "FATAL: error by parsing contained reads list: %s\n",
gt_error_get(err));
exit(EXIT_FAILURE);
}
gt_error_delete(err);
if (found_nofreads != expected_nofreads)
{
fprintf(stderr, "FATAL: error by parsing contained reads list: "
"file specifies a wrong number of reads\nexpected "GT_WU", found "
GT_WU"\n", expected_nofreads, found_nofreads);
exit(EXIT_FAILURE);
}
return gt_cntlist_count(*cntlist, found_nofreads);
}
static void gt_hpol_processor_output_segment(GtAlignedSegment *as,
bool may_be_gapped, GtFile *outfp, const char *desc)
{
unsigned long slen;
if (may_be_gapped)
gt_aligned_segment_ungap_seq_and_qual(as);
slen = (unsigned long)strlen(gt_aligned_segment_seq(as));
gt_assert(slen == (unsigned long)strlen(gt_aligned_segment_qual(as)));
if (gt_aligned_segment_is_reverse(as))
{
GtError *err = gt_error_new();
char *q = gt_aligned_segment_qual(as), tmp;
unsigned long i;
for (i = 0; i < (slen + 1UL) >> 1; i++)
{
tmp = q[i];
q[i] = q[slen - i - 1UL];
q[slen - i - 1UL] = tmp;
}
gt_assert((unsigned long)strlen(gt_aligned_segment_qual(as)) == slen);
if (gt_reverse_complement(gt_aligned_segment_seq(as), slen, err) != 0)
{
fprintf(stderr, "error: %s", gt_error_get(err));
exit(EXIT_FAILURE);
}
gt_error_delete(err);
}
gt_fastq_show_entry((desc != NULL) ? desc :
gt_aligned_segment_description(as), gt_aligned_segment_seq(as),
gt_aligned_segment_qual(as), slen, 0, false, outfp);
}
static int canvas_cairo_file_lua_new_generic(lua_State *L, GtGraphicsOutType t)
{
GtCanvas **canvas;
GtImageInfo **ii;
unsigned int width,
height;
GtError *err;
GtStyle *style;
width = luaL_checkint(L, 1);
height = luaL_checkint(L, 2);
/* create canvas */
style = gt_lua_get_style_from_registry(L);
canvas = lua_newuserdata(L, sizeof (GtCanvas*));
gt_assert(canvas);
/* if a imageinfo object is passed, it must be correct type */
if (lua_isnil(L, 3)) {
err = gt_error_new();
*canvas = gt_canvas_cairo_file_new(style, t, width, height, NULL, err);
} else {
ii = check_imageinfo(L, 3);
err = gt_error_new();
*canvas = gt_canvas_cairo_file_new(style, t, width, height, *ii, err);
}
if (gt_error_is_set(err))
return gt_lua_error(L, err);
gt_error_delete(err);
luaL_getmetatable(L, CANVAS_METATABLE);
lua_setmetatable(L, -2);
return 1;
}
static void proc_env_options(void)
{
int argc;
char *env_options, **argv;
GtSplitter *splitter;
GtError *err;
/* construct argument vector from $GT_ENV_OPTIONS */
env_options = getenv("GT_ENV_OPTIONS");
if (!env_options)
return;
env_options = gt_cstr_dup(env_options); /* make writeable copy */
splitter = gt_splitter_new();
gt_splitter_split(splitter, env_options, strlen(env_options), ' ');
argc = gt_splitter_size(splitter);
argv = gt_cstr_array_preprend((const char**) gt_splitter_get_tokens(splitter),
"env");
argc++;
/* parse options contained in $GT_ENV_OPTIONS */
err = gt_error_new();
switch (parse_env_options(argc, (const char**) argv, err)) {
case GT_OPTION_PARSER_OK: break;
case GT_OPTION_PARSER_ERROR:
fprintf(stderr, "error parsing $GT_ENV_OPTIONS: %s\n", gt_error_get(err));
gt_error_unset(err);
break;
case GT_OPTION_PARSER_REQUESTS_EXIT: break;
}
gt_error_delete(err);
gt_free(env_options);
gt_splitter_delete(splitter);
gt_cstr_array_delete(argv);
}
static void gv_test_calc_integrity(AgnUnitTest *test)
{
const char *filename = "data/gff3/gaeval-stream-unit-test-2.gff3";
GtNodeStream *align_in = gt_gff3_in_stream_new_unsorted(1, &filename);
AgnGaevalParams params = { 0.6, 0.3, 0.05, 0.05, 400, 200, 100 };
GtNodeVisitor *nv = agn_gaeval_visitor_new(align_in, params);
AgnGaevalVisitor *gv = gaeval_visitor_cast(nv);
gt_node_stream_delete(align_in);
GtNodeStream *gff3in = gt_gff3_in_stream_new_unsorted(1, &filename);
GtHashmap *typestokeep = gt_hashmap_new(GT_HASH_STRING, NULL, NULL);
gt_hashmap_add(typestokeep, "mRNA", "mRNA");
GtNodeStream *filtstream = agn_filter_stream_new(gff3in, typestokeep);
GtLogger *logger = gt_logger_new(true, "", stderr);
GtNodeStream *ics = agn_infer_cds_stream_new(filtstream, NULL, logger);
GtNodeStream *ies = agn_infer_exons_stream_new(ics, NULL, logger);
GtError *error = gt_error_new();
GtArray *feats = gt_array_new( sizeof(GtFeatureNode *) );
GtNodeStream *featstream = gt_array_out_stream_new(ies, feats, error);
int result = gt_node_stream_pull(featstream, error);
if(result == -1)
{
fprintf(stderr, "[AgnGaevalVisitor::gv_test_calc_integrity] error "
"processing GFF3: %s\n", gt_error_get(error));
return;
}
gt_node_stream_delete(gff3in);
gt_node_stream_delete(filtstream);
gt_node_stream_delete(featstream);
gt_node_stream_delete(ics);
gt_node_stream_delete(ies);
gt_logger_delete(logger);
gt_hashmap_delete(typestokeep);
agn_assert(gt_array_size(feats) == 2);
GtFeatureNode *g1 = *(GtFeatureNode **)gt_array_get(feats, 0);
GtFeatureNode *g2 = *(GtFeatureNode **)gt_array_get(feats, 1);
double cov1 = gaeval_visitor_calculate_coverage(gv, g1, error);
double cov2 = gaeval_visitor_calculate_coverage(gv, g2, error);
double int1 = gaeval_visitor_calculate_integrity(gv, g1, cov1, NULL, error);
double int2 = gaeval_visitor_calculate_integrity(gv, g2, cov2, NULL, error);
bool test1 = fabs(cov1 - 1.000) < 0.001 &&
fabs(cov2 - 0.997) < 0.001 &&
fabs(int1 - 0.850) < 0.001 &&
fabs(int2 - 0.863) < 0.001;
agn_unit_test_result(test, "calculate integrity", test1);
gt_error_delete(error);
gt_array_delete(feats);
gt_genome_node_delete((GtGenomeNode *)g1);
gt_genome_node_delete((GtGenomeNode *)g2);
gt_node_visitor_delete(nv);
}
int gt_lua_error(lua_State *L, GtError *err)
{
gt_assert(L && err);
gt_assert(gt_error_is_set(err));
luaL_where(L, 1);
lua_pushstring(L, gt_error_get(err));
gt_error_delete(err);
lua_concat(L, 2);
return lua_error(L);
}
static int encseq_lua_mirror(lua_State *L)
{
GtEncseq **encseq;
GtError *err = gt_error_new();
encseq = check_encseq(L, 1);
gt_assert(*encseq);
luaL_argcheck(L, !gt_encseq_is_mirrored(*encseq), 1, "is already mirrored");
if (gt_encseq_mirror(*encseq, err) != 0)
gt_lua_error(L, err);
gt_error_delete(err);
return 0;
}
static int feature_index_lua_get_range_for_seqid(lua_State *L)
{
GtFeatureIndex **feature_index;
const char *seqid;
GtError *err;
GtRange range;
bool has_seqid;
feature_index = check_feature_index(L, 1);
seqid = luaL_checkstring(L, 2);
err = gt_error_new();
if (gt_feature_index_has_seqid(*feature_index, &has_seqid, seqid, err))
return gt_lua_error(L, err);
gt_error_delete(err);
luaL_argcheck(L, has_seqid, 2,
"feature_index does not contain seqid");
err = gt_error_new();
if (gt_feature_index_get_range_for_seqid(*feature_index, &range, seqid, err))
return gt_lua_error(L, err);
gt_error_delete(err);
return gt_lua_range_push(L, range);
}
static int genome_node_lua_accept(lua_State *L)
{
GtGenomeNode **gn;
GtNodeVisitor **gv;
GtError *err;
gn = check_genome_node(L, 1);
gv = check_genome_visitor(L, 2);
err = gt_error_new();
if (gt_genome_node_accept(*gn, *gv, err))
return gt_lua_error(L, err);
gt_error_delete(err);
return 0;
}
int main(int argc, char *argv[])
{
GtError *err;
GtWord readlong;
bool haserr = false, withsequence = false;
char *indexname = NULL, *mumlength_string = NULL;
if (argc == 3)
{
withsequence = false;
indexname = argv[1];
mumlength_string = argv[2];
} else
{
if (argc == 4)
{
if (strcmp(argv[1],"-s") != 0)
{
usage(argv[0]);
return EXIT_FAILURE;
}
withsequence = true;
indexname = argv[2];
mumlength_string = argv[3];
} else
{
usage(argv[0]);
return EXIT_FAILURE;
}
}
if (sscanf(mumlength_string,GT_WD,&readlong) != 1 || readlong < 0)
{
usage(argv[0]);
return EXIT_FAILURE;
}
gt_lib_init();
err = gt_error_new();
if (gt_stree_mum(indexname, (GtUword) readlong, withsequence, err) != 0)
{
fprintf(stderr,"%s: failure when enumerating branch locs for "
" index %s: %s\n",
argv[0],indexname,gt_error_get(err));
haserr = true;
}
gt_error_delete(err);
if (gt_lib_clean() != 0)
{
return GT_EXIT_PROGRAMMING_ERROR;
}
return haserr ? EXIT_FAILURE : EXIT_SUCCESS;
}
static int feature_index_lua_add_gff3file(lua_State *L)
{
GtFeatureIndex **fi;
const char *filename;
GtError *err;
gt_assert(L);
fi = check_feature_index(L, 1);
filename = luaL_checkstring(L, 2);
err = gt_error_new();
if (gt_feature_index_add_gff3file(*fi, filename, err))
return gt_lua_error(L, err);
gt_error_delete(err);
return 0;
}
static int encseq_encoder_lua_use_representation(lua_State *L)
{
GtEncseqEncoder **encoder;
const char *repr;
GtError *err;
encoder = check_encseq_encoder(L, 1);
repr = luaL_checkstring(L, 2);
gt_assert(*encoder);
err = gt_error_new();
if (gt_encseq_encoder_use_representation(*encoder, repr, err) != 0)
gt_lua_error(L, err);
gt_error_delete(err);
return 0;
}
static int encseq_encoder_lua_use_symbolmap_file(lua_State *L)
{
GtEncseqEncoder **encoder;
const char *smapfile;
GtError *err;
encoder = check_encseq_encoder(L, 1);
smapfile = luaL_checkstring(L, 2);
gt_assert(*encoder);
err = gt_error_new();
if (gt_encseq_encoder_use_symbolmap_file(*encoder, smapfile, err) != 0)
gt_lua_error(L, err);
gt_error_delete(err);
return 0;
}
static int gt_node_stream_lua_next_tree(lua_State *L)
{
GtNodeStream **gs = check_genome_stream(L, 1);
GtGenomeNode *gn;
GtError *err = gt_error_new();
if (gt_node_stream_next(*gs, &gn, err))
return gt_lua_error(L, err); /* handle error */
else if (gn)
gt_lua_genome_node_push(L, gn);
else
lua_pushnil(L);
gt_error_delete(err);
return 1;
}
static int layout_lua_get_height(lua_State *L)
{
GtLayout **layout;
GtError *err;
int had_err = 0;
unsigned long height;
layout = check_layout(L, 1);
err = gt_error_new();
had_err = gt_layout_get_height(*layout, &height, err);
lua_pushnumber(L, height);
if (had_err < 0)
return gt_lua_error(L, err);
gt_error_delete(err);
return 1;
}
static int layout_lua_sketch(lua_State *L)
{
GtLayout **layout;
GtCanvas **canvas;
GtError *err;
int had_err = 0;
layout = check_layout(L, 1);
canvas = check_canvas(L, 2);
err = gt_error_new();
had_err = gt_layout_sketch(*layout, *canvas, err);
if (had_err < 0)
return gt_lua_error(L, err);
gt_error_delete(err);
return 0;
}
static int encseq_builder_lua_build(lua_State *L)
{
GtEncseqBuilder **builder;
GtEncseq *encseq;
GtError *err;
builder = check_encseq_builder(L, 1);
err = gt_error_new();
encseq = gt_encseq_builder_build(*builder, err);
if (encseq == NULL) {
gt_lua_error(L, err);
} else
gt_lua_encseq_push(L, encseq);
gt_error_delete(err);
return 1;
}
static int feature_index_lua_get_seqids(lua_State *L)
{
GtFeatureIndex **feature_index;
GtStrArray *seqids;
GtError *err;
feature_index = check_feature_index(L, 1);
err = gt_error_new();
seqids = gt_feature_index_get_seqids(*feature_index, err);
if (!seqids)
return gt_lua_error(L, err);
gt_error_delete(err);
/* push table containing sequence ids onto the stack */
gt_lua_push_strarray_as_table(L, seqids);
gt_str_array_delete(seqids);
return 1;
}
static int feature_index_lua_add_region_node(lua_State *L)
{
GtFeatureIndex **fi;
GtGenomeNode **gn;
GtRegionNode *rn;
GtError *err;
gt_assert(L);
fi = check_feature_index(L, 1);
gn = check_genome_node(L, 2);
rn = gt_region_node_try_cast(*gn);
luaL_argcheck(L, rn, 2, "not a region node");
err = gt_error_new();
if (gt_feature_index_add_region_node(*fi, rn, err))
return gt_lua_error(L, err);
gt_error_delete(err);
return 0;
}
static int feature_index_lua_has_seqid(lua_State *L)
{
GtFeatureIndex **feature_index;
bool has_seqid;
const char *seqid;
GtError *err;
int had_err = 0;
feature_index = check_feature_index(L, 1);
seqid = luaL_checkstring(L, 2);
err = gt_error_new();
had_err = gt_feature_index_has_seqid(*feature_index, &has_seqid, seqid, err);
if (had_err)
return gt_lua_error(L, err);
gt_error_delete(err);
lua_pushboolean(L, has_seqid);
return 1;
}
static int encseq_loader_lua_load(lua_State *L)
{
GtEncseqLoader **loader;
const char *indexname;
GtError *err = gt_error_new();
GtEncseq *encseq;
loader = check_encseq_loader(L, 1);
indexname = luaL_checkstring(L, 2);
gt_assert(*loader);
encseq = gt_encseq_loader_load(*loader, indexname, err);
if (encseq == NULL)
gt_lua_error(L, err);
else {
gt_lua_encseq_push(L, encseq);
}
gt_error_delete(err);
return 1;
}
static int feature_index_lua_get_features_for_seqid(lua_State *L)
{
GtFeatureIndex **feature_index;
const char *seqid;
GtArray *features;
GtError *err;
feature_index = check_feature_index(L, 1);
seqid = luaL_checkstring(L, 2);
err = gt_error_new();
features = gt_feature_index_get_features_for_seqid(*feature_index, seqid,
err);
if (!features)
return gt_lua_error(L, err);
gt_error_delete(err);
push_features_as_table(L, features);
gt_array_delete(features);
return 1;
}
static int feature_index_lua_get_first_seqid(lua_State *L)
{
GtFeatureIndex **feature_index;
char *seqid;
GtError *err;
feature_index = check_feature_index(L, 1);
err = gt_error_new();
seqid = gt_feature_index_get_first_seqid(*feature_index, err);
if (gt_error_is_set(err))
return gt_lua_error(L, err);
gt_error_delete(err);
if (seqid) {
lua_pushstring(L, seqid);
gt_free(seqid);
} else
lua_pushnil(L);
return 1;
}
