cstrittmatter@0: """Entry script"""
cstrittmatter@0: 
cstrittmatter@0: __version__ = '1.0.0'
cstrittmatter@0: import argparse, datetime, json, logging, os, pandas, sys
cstrittmatter@0: import lexmapr.pipeline, lexmapr.run_summary
cstrittmatter@0: from lexmapr.definitions import arg_bins
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: def valid_input(file_path):
cstrittmatter@0:     '''Exits if input file is invalid'''
cstrittmatter@0:     _, file_ext = os.path.splitext(file_path)
cstrittmatter@0:     if file_ext.lower() != '.csv' and file_ext.lower() != '.tsv':
cstrittmatter@0:         sys.exit('Please supply a CSV or TSV input file with the correct file extension')
cstrittmatter@0:     if not os.path.exists(file_path):
cstrittmatter@0:         sys.exit(f'Input file named \"{file_path}\" not found')
cstrittmatter@0:     return(file_path.strip())
cstrittmatter@0: 
cstrittmatter@0: def valid_json(file_path):
cstrittmatter@0:     '''Outputs read JSON file and exits if file is invalid'''
cstrittmatter@0:     try:
cstrittmatter@0:         with open(file_path, 'r') as JSON_file:
cstrittmatter@0:             try:
cstrittmatter@0:                 return(json.load(JSON_file))
cstrittmatter@0:             except(json.decoder.JSONDecodeError):
cstrittmatter@0:                 sys.exit(f'User-defined bins not in readable JSON format')            
cstrittmatter@0:     except(FileNotFoundError):
cstrittmatter@0:         sys.exit(f'File named \"{file_path}\" not found')
cstrittmatter@0: 
cstrittmatter@0: def valid_list(list_str):
cstrittmatter@0:     '''Return list of user-defined ontologies'''
cstrittmatter@0:     return([x.strip().upper() for x in list_str.split(',')])
cstrittmatter@0: 
cstrittmatter@0: if __name__ == "__main__":
cstrittmatter@0:     # Parse arguments, initiate log file and start run
cstrittmatter@0:     arg_parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter)
cstrittmatter@0:     arg_parser.add_argument('input', help='input CSV or TSV file; required', type=valid_input)
cstrittmatter@0:     arg_parser.add_argument('-o', '--output', metavar='\b',
cstrittmatter@0:                             help='    output TSV file path; default is stdout')
cstrittmatter@0:     arg_parser.add_argument('-a', '--no_ancestors', action='store_true',
cstrittmatter@0:                             help='remove ancestral terms from output')
cstrittmatter@0:     arg_parser.add_argument('-b', '--bin', action='store_true',
cstrittmatter@0:                             help='classify samples into default bins')
cstrittmatter@0:     arg_parser.add_argument('-e', '--embl_ontol', metavar='\b', type=valid_list,
cstrittmatter@0:                             help='    user-defined comma-separated ontology short names')
cstrittmatter@0:     arg_parser.add_argument('-f', '--full', action='store_true', help='full output format')
cstrittmatter@0:     arg_parser.add_argument('-g', '--graph', action='store_true',
cstrittmatter@0:                             help='visualize summaries of mapping and binning')
cstrittmatter@0:     arg_parser.add_argument('-j', '--graph_only', action='store_true',
cstrittmatter@0:                             help='only perform visualization with LexMapr output')
cstrittmatter@0:     arg_parser.add_argument('-r', '--remake_cache', action='store_true',
cstrittmatter@0:                             help='remake cached resources')
cstrittmatter@0:     arg_parser.add_argument('-u', '--user_bin', metavar='\b', type=valid_json,
cstrittmatter@0:                             help='    path to JSON file with user-defined bins')
cstrittmatter@0:     arg_parser.add_argument('-v', '--version', action='version',
cstrittmatter@0:                             version='%(prog)s '+__version__)
cstrittmatter@0: 
cstrittmatter@0:     # TODO: encoding argument addded to logging.basicConfig in Python 3.9; now defaults to open()
cstrittmatter@0:     run_args = arg_parser.parse_args()
cstrittmatter@0:     if run_args.user_bin is not None:
cstrittmatter@0:         run_args.bin = True
cstrittmatter@0:         arg_bins = run_args.user_bin
cstrittmatter@0: 
cstrittmatter@0:     logging.basicConfig(filename='lexmapr_run.log', level=logging.DEBUG)
cstrittmatter@0: 
cstrittmatter@0:     if run_args.graph_only:
cstrittmatter@0:         try:
cstrittmatter@0:             mapping_results = pandas.read_csv(run_args.input, delimiter='\t')
cstrittmatter@0:         except:
cstrittmatter@0:             sys.exit('Input file not readable or not in expected format')
cstrittmatter@0:         needed_columns = ['Matched_Components','Match_Status (Macro Level)']+list(arg_bins.keys())
cstrittmatter@0:         missing_columns = set(needed_columns).difference(set(mapping_results.columns))
cstrittmatter@0:         if missing_columns:
cstrittmatter@0:             sys.exit(f'Missing column(s) {missing_columns} from input file')
cstrittmatter@0:         t0 = datetime.datetime.now()
cstrittmatter@0:         logging.info(f'Run start: {t0}')
cstrittmatter@0:         logging.info('Graphing only')
cstrittmatter@0:         print('\nGraphing only...')
cstrittmatter@0:         lexmapr.run_summary.figure_folder()
cstrittmatter@0:         lexmapr.run_summary.report_results(run_args.input, list(arg_bins.keys()))
cstrittmatter@0:         lexmapr.run_summary.visualize_results(run_args.input, list(arg_bins.keys()))
cstrittmatter@0:         print('\t'+f'Done! {datetime.datetime.now()-t0} passed'.ljust(60)+'\n')
cstrittmatter@0:     else:
cstrittmatter@0:         logging.info(f'Run start: {datetime.datetime.now()}')
cstrittmatter@0:         lexmapr.pipeline.run(run_args)
cstrittmatter@0: 
cstrittmatter@0:     logging.info(f'Run end: {datetime.datetime.now()}\n')