def processPrediXcanFiles(self):
logging.info("Loading people")
all_people = Person.Person.loadPeople(self.samples_input, '\t', False)
selected_people = Person.Person.loadPeople(self.samples_output)
selected_people_by_id = {p.id: p for p in selected_people}
logging.info("%d total people, %d selected", len(all_people),
len(selected_people_by_id))
logging.info("Loading snps")
snp_data_set = DataSet.DataSetFileUtilities.loadFromCompressedFile(
self.snp_list)
snp_dict = {k: True for k in snp_data_set.data}
print len(snp_dict.keys())
contents = Utilities.contentsWithPatternsFromFolder(
self.dosage_folder, ["dosage.txt.gz"])
for content_name in contents:
input_path = os.path.join(self.dosage_folder, content_name)
fileBuilder = PrediXcanFormatUtilities.PrediXcanFormatFilteredFilesProcess(
input_path, self.output_folder, content_name, all_people,
selected_people_by_id, snp_dict)
if self.output_format == Formats.IMPUTE:
fileBuilder.buildIMPUTE()
if self.output_format == Formats.PrediXcan:
fileBuilder.buildPrediXcan()
else:
raise Exceptions.InvalidOutputFormat(self.output_format)
def processIMPUTEFiles(self):
logging.info("Loading people")
names = Utilities.hapNamesFromFolder(self.dosage_folder)
all_people = Person.Person.loadPeople(self.samples_input)
selected_people = Person.Person.loadPeople(self.samples_output,
delim=" ")
selected_people_by_id = {p.id: p for p in selected_people}
logging.info("Loading snps")
snp_data_set = DataSet.DataSetFileUtilities.loadFromCompressedFile(
self.snp_list)
snp_dict = {rsid: True for rsid in snp_data_set.data}
for name in names:
output = os.path.join(self.output_folder, name)
filter = ThousandGenomesUtilities.IMPUTEFilteredDosageFileBuilder()
filter.base_path = self.dosage_folder
filter.name = name
filter.output_pattern = output
filter.snp_dict = snp_dict
filter.all_people = all_people
filter.selected_people_by_id = selected_people_by_id
if self.output_format == Formats.IMPUTE:
filter.buildIMPUTE()
elif self.output_format == Formats.PrediXcan:
search = self.chromosome_in_name_regex.search(name)
exitIf(search is None, Exceptions.InvalidInputFormat, \
"No files found in '%s' that match the pattern, '%s'" \
% (self.dosage_folder, self.chromosome_in_name_regex.pattern))
chr = search.group(1)
filter.chromosome_name = chr
filter.buildPrediXcan()
else:
raise Exceptions.InvalidOutputFormat(self.output_format)
