Permalink
Cannot retrieve contributors at this time
Fetching contributors…
Cannot retrieve contributors at this time
| #!/usr/bin/python | |
| import stat | |
| import os | |
| import sys | |
| import time | |
| def main(): | |
| base_name = sys.argv[1] | |
| max_seq_file = 20000 | |
| max_files_dir = 1000 | |
| out_dir = "." | |
| count_seq = 0 | |
| count_seq_file = 0 | |
| count_files_dir = 0 | |
| outf = None | |
| dir_id = 0 | |
| file_id = 0 | |
| outf_filename = "" | |
| perf_start = time.time() | |
| try: | |
| os.makedirs("%s/%06d" %(out_dir, dir_id)) | |
| except: | |
| pass | |
| outf_filename = "%s/%06d/%s-%06d-%06d.fastq" \ | |
| %(out_dir, dir_id, base_name, dir_id, file_id) | |
| outf = open(outf_filename, "w", buffering=500*(1024**2)) | |
| line_count = 0 | |
| for line in sys.stdin: | |
| line_count += 1 | |
| # sanity check | |
| if line_count == 1 and line[0] != '@': | |
| print("Error: sequence number %d does not start with @" %(count_seq)) | |
| sys.exit(1) | |
| outf.write(line) | |
| if line_count == 4: | |
| # we have a full sequence | |
| count_seq += 1 | |
| count_seq_file += 1 | |
| line_count = 0 | |
| if count_seq_file == max_seq_file: | |
| count_files_dir += 1 | |
| if count_files_dir == max_files_dir: | |
| dir_id += 1 | |
| file_id = 0 | |
| count_files_dir = 0 | |
| try: | |
| os.makedirs("%s/%06d" %(out_dir, dir_id)) | |
| except: | |
| pass | |
| else: | |
| file_id += 1 | |
| count_seq_file = 0 | |
| outf.close() | |
| outf_filename = "%s/%06d/%s-%06d-%06d.fastq" \ | |
| %(out_dir, dir_id, base_name, dir_id, file_id) | |
| outf = open(outf_filename, "w", buffering=500*(1024**2)) | |
| outf.close() | |
| # corner case where outf is empty at this point | |
| if count_seq_file == 0: | |
| try: | |
| os.unlink(outf_filename) | |
| except: | |
| pass | |
| delta = time.time() - perf_start | |
| print("Total: %d sequences processed. %.0f sequences / s" %(count_seq, count_seq / delta)) | |
| if __name__ == "__main__": | |
| main() | |