def main(argv=None):
# Parse command line
try:
settings, args = process_command_line(None)
except BaseException:
return 1
# Parse format config file
try:
rows = read_config_file_as_table(settings.format_config)
our_file_list = [row["unified_path"] for row in rows]
except ConfigException as ex:
print ex.message
return 1
# Used database name "article_refactor_24_3_2016"
header, final_table = merge_results(our_tables, their_tables,
settings.threshold, settings.database)
# Generate non formatted table
output_path = os.path.join(settings.workdir, settings.output_file)
temp_file_path = output_path + "_unformmated"
with open(temp_file_path, "wb") as fl:
fl.write("%s\n" % "\t".join(header))
for row in final_table:
fl.write("%s\n" % "\t".join(str(val) for val in row))
# Format the temporary table
format_final_table(temp_file_path, our_file_list, output_path)
return 0
return settings, args def upload_tables(table_list): for our_file, table_name in table_list: print "*" * 100 print "Uploading %s with name %s" % (our_file, table_name) upload_table(our_file, table_name) print "*" * 100 if __name__ == "__main__": settings, args = process_command_line(None) try: config_rows = read_config_file_as_table(settings.config_file) except ConfigException as ex: print ex exit(1) table_list = zip([row["file_name"] for row in config_rows], [row["table_name"] for row in config_rows]) upload_tables(table_list) exit(0)
return settings, args
def upload_tables(table_list):
for our_file, table_name in table_list:
print "*" * 100
print "Uploading %s with name %s" % (our_file, table_name)
upload_table(our_file, table_name)
print "*" * 100
if __name__ == "__main__":
settings, args = process_command_line(None)
try:
config_rows = read_config_file_as_table(settings.config_file)
except ConfigException as ex:
print ex
exit(1)
table_list = zip([row["file_name"] for row in config_rows],
[row["table_name"] for row in config_rows])
upload_tables(table_list)
exit(0)
def run():
settings, args = process_command_line(None)
try:
config_rows = read_config_file_as_table(settings.config_file)
if settings.unified_config_file:
unified_config_rows = read_config_file_as_table(
settings.unified_config_file)
except ConfigException as ex:
print ex
return 1
interaction_type_to_condition_files_dictionary = get_condition_files(
config_rows)
signif_dictionary = generate_signif_per_conditions_dictionary(config_rows)
# max_lib_dictionary = generate_max_lib_per_condition_dictionary(config_rows)
# Mark whether option was set
categories_lists = {}
# categories_lists = {DIV_OPTION_SIGNIF: signif_dictionary}
if settings.signif:
categories_lists[DIV_OPTION_SIGNIF] = signif_dictionary
if settings.known_targets_file:
known_pair_list = generate_known_pairs_list(
settings.known_targets_file)
categories_lists[DIV_OPTION_KNOWN_TARGETS] = known_pair_list
if settings.no_binding_file:
binding_pair_list = generate_binding_pairs_list(
settings.no_binding_file)
categories_lists[DIV_OPTION_BINDING] = binding_pair_list
if settings.questionable_file:
questionable_pair_list = generate_questionable_pairs_list(
settings.questionable_file)
categories_lists[DIV_OPTION_QUESTIONABLE] = questionable_pair_list
interaction_to_condition_totals = {}
interaction_to_condition_libs = {}
interaction_to_condition_signif_libs = {}
# Go over the files and make the total count of interactions per interaction type
for interaction_type, condition_files_dictionary in interaction_type_to_condition_files_dictionary.items(
):
interaction_to_condition_totals[interaction_type] = {}
interaction_to_condition_libs[interaction_type] = {}
interaction_to_condition_signif_libs[interaction_type] = {}
for condition_name, file_list in condition_files_dictionary.items():
interaction_to_condition_totals[interaction_type][
condition_name] = {}
interaction_to_condition_libs[interaction_type][
condition_name] = {}
interaction_to_condition_signif_libs[interaction_type][
condition_name] = {}
for file_path in file_list:
fp = FileProcessor(
file_path, {
"read_count": CountHandler(),
"signif_lib_count": MaxLibHandler()
})
fp.process()
interactions_dct = fp.row_handlers["read_count"].dct
max_signif_lib_dct = fp.row_handlers["signif_lib_count"].dct
# Handle total count
for name, count in interactions_dct.items():
# Add the name for the first time
if name not in interaction_to_condition_totals[
interaction_type][condition_name]:
interaction_to_condition_totals[interaction_type][
condition_name][name] = 0
# Increase the count
interaction_to_condition_totals[interaction_type][
condition_name][name] += count
# Handle signif libs
for name, lib_count in max_signif_lib_dct.items():
# Add the name for the first time
if name not in interaction_to_condition_signif_libs[
interaction_type][condition_name]:
interaction_to_condition_signif_libs[interaction_type][
condition_name][name] = 0
# Set the maximal lib
interaction_to_condition_signif_libs[interaction_type][condition_name][name] = \
max(interaction_to_condition_signif_libs[interaction_type][condition_name][name], lib_count)
# Handle total libs
for name, lib_count in interactions_dct.items():
# Add the name for the first time
if name not in interaction_to_condition_libs[
interaction_type][condition_name]:
interaction_to_condition_libs[interaction_type][
condition_name][name] = 0
# Set the maximal lib
interaction_to_condition_libs[interaction_type][
condition_name][name] += 1
# Only if unified config file was inserted
if settings.unified_config_file:
unified_interaction_type_to_condition_files_dictionary = get_condition_files(
unified_config_rows)
unified_interaction_to_condition_totals = {}
unified_interaction_to_condition_libs = {}
for interaction_type, condition_files_dictionary in unified_interaction_type_to_condition_files_dictionary.items(
):
unified_interaction_to_condition_totals[interaction_type] = {}
unified_interaction_to_condition_libs[interaction_type] = {}
for condition_name, file_list in condition_files_dictionary.items(
):
unified_interaction_to_condition_totals[interaction_type][
condition_name] = {}
for file_path in file_list:
fp = FileProcessor(
file_path, {
"lib_count":
UnifiedLibHandler(
interaction_to_condition_libs[interaction_type]
[condition_name])
})
fp.process()
unified_interaction_to_condition_libs[interaction_type][condition_name] = \
fp.row_handlers["lib_count"].dct
# Generate group tables per condition
for interaction_type, condition_totals in interaction_to_condition_totals.items(
):
perms = []
# Generate all permutations for the flags
for i in range(len(categories_lists) + 1):
values = [True] * i + [False] * (len(categories_lists) - i)
perms += list(set(itertools.permutations(values)))
# Generate the different groups according to t
for permutation in perms:
categories_results = {
category: permutation[i]
for i, category in enumerate(categories_lists.keys())
}
generate_table(
interaction_type, condition_totals,
interaction_to_condition_libs[interaction_type],
interaction_to_condition_signif_libs[interaction_type],
categories_results, categories_lists)
# Generate unified tables
if settings.unified_config_file:
for interaction_type, condition_to_libs in unified_interaction_to_condition_libs.items(
):
for condition_name, names_to_libs in condition_to_libs.items():
with open(
"output/unified_%s_%s.table" %
(interaction_type, condition_name), "wb") as fl:
writer = csv.writer(fl,
delimiter='\t',
quotechar='|',
quoting=csv.QUOTE_MINIMAL)
writer.writerow(["libs", "known", "unknown"])
libs_max = max(
set([int(val) for key, val in names_to_libs.items()]))
for i in range(libs_max + 1):
writer.writerow([
i,
sum([
1 for name, val in names_to_libs.items()
if val == i and is_known(
name,
categories_lists[DIV_OPTION_KNOWN_TARGETS])
]),
sum([
1 for name, val in names_to_libs.items()
if val == i and not is_known(
name,
categories_lists[DIV_OPTION_KNOWN_TARGETS])
])
])
return 0
def run():
settings, args = process_command_line(None)
try:
config_rows = read_config_file_as_table(settings.config_file)
if settings.unified_config_file:
unified_config_rows = read_config_file_as_table(settings.unified_config_file)
except ConfigException as ex:
print ex
return 1
interaction_type_to_condition_files_dictionary = get_condition_files(config_rows)
signif_dictionary = generate_signif_per_conditions_dictionary(config_rows)
# max_lib_dictionary = generate_max_lib_per_condition_dictionary(config_rows)
# Mark whether option was set
categories_lists = {}
# categories_lists = {DIV_OPTION_SIGNIF: signif_dictionary}
if settings.signif:
categories_lists[DIV_OPTION_SIGNIF] = signif_dictionary
if settings.known_targets_file:
known_pair_list = generate_known_pairs_list(settings.known_targets_file)
categories_lists[DIV_OPTION_KNOWN_TARGETS] = known_pair_list
if settings.no_binding_file:
binding_pair_list = generate_binding_pairs_list(settings.no_binding_file)
categories_lists[DIV_OPTION_BINDING] = binding_pair_list
if settings.questionable_file:
questionable_pair_list = generate_questionable_pairs_list(settings.questionable_file)
categories_lists[DIV_OPTION_QUESTIONABLE] = questionable_pair_list
interaction_to_condition_totals = {}
interaction_to_condition_libs = {}
interaction_to_condition_signif_libs = {}
# Go over the files and make the total count of interactions per interaction type
for interaction_type, condition_files_dictionary in interaction_type_to_condition_files_dictionary.items():
interaction_to_condition_totals[interaction_type] = {}
interaction_to_condition_libs[interaction_type] = {}
interaction_to_condition_signif_libs[interaction_type] = {}
for condition_name, file_list in condition_files_dictionary.items():
interaction_to_condition_totals[interaction_type][condition_name] = {}
interaction_to_condition_libs[interaction_type][condition_name] = {}
interaction_to_condition_signif_libs[interaction_type][condition_name] = {}
for file_path in file_list:
fp = FileProcessor(file_path, {"read_count": CountHandler(),
"signif_lib_count": MaxLibHandler()})
fp.process()
interactions_dct = fp.row_handlers["read_count"].dct
max_signif_lib_dct = fp.row_handlers["signif_lib_count"].dct
# Handle total count
for name, count in interactions_dct.items():
# Add the name for the first time
if name not in interaction_to_condition_totals[interaction_type][condition_name]:
interaction_to_condition_totals[interaction_type][condition_name][name] = 0
# Increase the count
interaction_to_condition_totals[interaction_type][condition_name][name] += count
# Handle signif libs
for name, lib_count in max_signif_lib_dct.items():
# Add the name for the first time
if name not in interaction_to_condition_signif_libs[interaction_type][condition_name]:
interaction_to_condition_signif_libs[interaction_type][condition_name][name] = 0
# Set the maximal lib
interaction_to_condition_signif_libs[interaction_type][condition_name][name] = \
max(interaction_to_condition_signif_libs[interaction_type][condition_name][name], lib_count)
# Handle total libs
for name, lib_count in interactions_dct.items():
# Add the name for the first time
if name not in interaction_to_condition_libs[interaction_type][condition_name]:
interaction_to_condition_libs[interaction_type][condition_name][name] = 0
# Set the maximal lib
interaction_to_condition_libs[interaction_type][condition_name][name] += 1
# Only if unified config file was inserted
if settings.unified_config_file:
unified_interaction_type_to_condition_files_dictionary = get_condition_files(unified_config_rows)
unified_interaction_to_condition_totals = {}
unified_interaction_to_condition_libs = {}
for interaction_type, condition_files_dictionary in unified_interaction_type_to_condition_files_dictionary.items():
unified_interaction_to_condition_totals[interaction_type] = {}
unified_interaction_to_condition_libs[interaction_type] = {}
for condition_name, file_list in condition_files_dictionary.items():
unified_interaction_to_condition_totals[interaction_type][condition_name] = {}
for file_path in file_list:
fp = FileProcessor(file_path, {"lib_count": UnifiedLibHandler(interaction_to_condition_libs[interaction_type][condition_name])})
fp.process()
unified_interaction_to_condition_libs[interaction_type][condition_name] = \
fp.row_handlers["lib_count"].dct
# Generate group tables per condition
for interaction_type, condition_totals in interaction_to_condition_totals.items():
perms = []
# Generate all permutations for the flags
for i in range(len(categories_lists) + 1):
values = [True] * i + [False] * (len(categories_lists) - i)
perms += list(set(itertools.permutations(values)))
# Generate the different groups according to t
for permutation in perms:
categories_results = {category: permutation[i] for i, category in enumerate(categories_lists.keys())}
generate_table(interaction_type,
condition_totals,
interaction_to_condition_libs[interaction_type],
interaction_to_condition_signif_libs[interaction_type],
categories_results,
categories_lists)
# Generate unified tables
if settings.unified_config_file:
for interaction_type, condition_to_libs in unified_interaction_to_condition_libs.items():
for condition_name, names_to_libs in condition_to_libs.items():
with open("output/unified_%s_%s.table" % (interaction_type, condition_name), "wb") as fl:
writer = csv.writer(fl, delimiter='\t', quotechar='|', quoting=csv.QUOTE_MINIMAL)
writer.writerow(["libs", "known", "unknown"])
libs_max = max(set([int(val) for key, val in names_to_libs.items()]))
for i in range(libs_max + 1):
writer.writerow([i,
sum([1 for name, val in names_to_libs.items() if val == i and is_known(name, categories_lists[DIV_OPTION_KNOWN_TARGETS])]),
sum([1 for name, val in names_to_libs.items() if val == i and not is_known(name, categories_lists[DIV_OPTION_KNOWN_TARGETS])])])
return 0
