def filter_transcripts(classify_dir, min_prec, min_rec, min_spec,
opt_variable, tmp_dir):
# setup input and output files
lib_counts_file = os.path.join(classify_dir, LIB_COUNTS_FILE)
lib_counts_list = list(LibCounts.from_file(lib_counts_file))
# filter each category
expr_gtf_files = []
bkgd_gtf_files = []
for category_key in CATEGORIES:
category_str = category_int_to_str[category_key]
category_lib_counts = [x for x in lib_counts_list
if x.category_counts[category_key] > 0]
library_ids = [x.library_id for x in category_lib_counts]
cinfo = CategoryInfo.create(library_ids, category_key,
category_str, classify_dir)
filter_category(category_lib_counts, cinfo, min_prec, min_rec,
min_spec, opt_variable, tmp_dir)
expr_gtf_files.append(cinfo.unann_expr_gtf_file)
bkgd_gtf_files.append(cinfo.unann_bkgd_gtf_file)
# only need one set of annotated gtf files
expr_gtf_files.extend([cinfo.ann_expr_gtf_file,
cinfo.ann_bkgd_gtf_file])
# merge transcripts
logging.info("Merging filtered transcripts")
expr_gtf_file = os.path.join(classify_dir, EXPR_GTF_FILE)
bkgd_gtf_file = os.path.join(classify_dir, BKGD_GTF_FILE)
merge_sort_gtf_files(expr_gtf_files, expr_gtf_file, tmp_dir=tmp_dir)
merge_sort_gtf_files(bkgd_gtf_files, bkgd_gtf_file, tmp_dir=tmp_dir)
def classify_transcripts(classify_dir, num_processors, gtf_score_attr,
tmp_dir):
# setup input and output files
lib_counts_file = os.path.join(classify_dir, LIB_COUNTS_FILE)
lib_counts_list = list(LibCounts.from_file(lib_counts_file))
library_ids = [x.library_id for x in lib_counts_list]
category_info_dict = {}
for category_key in CATEGORIES:
category_str = category_int_to_str[category_key]
cinfo = CategoryInfo.create(library_ids, category_key, category_str,
classify_dir)
category_info_dict[category_key] = cinfo
# write input files for classifier
logging.info("Writing classification input files category='%s'" %
(cinfo.category_str))
for transcripts in parse_gtf(open(cinfo.output_gtf_file)):
for t in transcripts:
# set transcript score
t.score = float(t.attrs.get(gtf_score_attr, 0.0))
library_id = t.attrs[GTFAttr.LIBRARY_ID]
fields = get_classification_fields(t)
# lookup file handle and open new file if necessary
if not library_id in cinfo.result_fh_dict:
cinfo.result_fh_dict[library_id] = open(
cinfo.result_file_dict[library_id], "w")
print >> cinfo.result_fh_dict[library_id], '\t'.join(
get_classification_header())
# write to file
print >> cinfo.result_fh_dict[library_id], '\t'.join(
map(str, fields))
# close open file handles
for fh in cinfo.result_fh_dict.itervalues():
fh.close()
for category_key, cinfo in category_info_dict.iteritems():
classify_tasks = []
for lib_counts in lib_counts_list:
# see if can run classifier on this file
if lib_counts.category_counts[category_key] > 0:
filename = cinfo.result_file_dict[lib_counts.library_id]
classify_tasks.append((lib_counts.library_id, filename))
# run classification
logging.info("Classifying transcripts category='%s'" %
(cinfo.category_str))
classify_category(cinfo, classify_tasks, num_processors, tmp_dir)
# sort results
sort_classification_results(cinfo.ctree_file, cinfo.sorted_ctree_file,
tmp_dir)
os.remove(cinfo.ctree_file)
def classify_transcripts(classify_dir, num_processors, gtf_score_attr,
tmp_dir):
# setup input and output files
lib_counts_file = os.path.join(classify_dir, LIB_COUNTS_FILE)
lib_counts_list = list(LibCounts.from_file(lib_counts_file))
library_ids = [x.library_id for x in lib_counts_list]
category_info_dict = {}
for category_key in CATEGORIES:
category_str = category_int_to_str[category_key]
cinfo = CategoryInfo.create(library_ids, category_key,
category_str, classify_dir)
category_info_dict[category_key] = cinfo
# write input files for classifier
logging.info("Writing classification input files category='%s'" % (cinfo.category_str))
for transcripts in parse_gtf(open(cinfo.output_gtf_file)):
for t in transcripts:
# set transcript score
t.score = float(t.attrs.get(gtf_score_attr, 0.0))
library_id = t.attrs[GTFAttr.LIBRARY_ID]
fields = get_classification_fields(t)
# lookup file handle and open new file if necessary
if not library_id in cinfo.result_fh_dict:
cinfo.result_fh_dict[library_id] = open(cinfo.result_file_dict[library_id], "w")
print >>cinfo.result_fh_dict[library_id], '\t'.join(get_classification_header())
# write to file
print >>cinfo.result_fh_dict[library_id], '\t'.join(map(str, fields))
# close open file handles
for fh in cinfo.result_fh_dict.itervalues():
fh.close()
for category_key, cinfo in category_info_dict.iteritems():
classify_tasks = []
for lib_counts in lib_counts_list:
# see if can run classifier on this file
if lib_counts.category_counts[category_key] > 0:
filename = cinfo.result_file_dict[lib_counts.library_id]
classify_tasks.append((lib_counts.library_id, filename))
# run classification
logging.info("Classifying transcripts category='%s'" % (cinfo.category_str))
classify_category(cinfo, classify_tasks, num_processors, tmp_dir)
# sort results
sort_classification_results(cinfo.ctree_file, cinfo.sorted_ctree_file, tmp_dir)
os.remove(cinfo.ctree_file)
