| Previous changeset 45:7aa966c488a4 (2020-01-22) Next changeset 47:3af9d394367c (2020-02-19) |
|
Commit message:
Uploaded |
|
added:
Marea/ras_generator.py |
| b |
| diff -r 7aa966c488a4 -r 5d5d01ef1d68 Marea/ras_generator.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/Marea/ras_generator.py Wed Jan 22 11:50:54 2020 -0500 |
| [ |
| b'@@ -0,0 +1,869 @@\n+from __future__ import division\n+import sys\n+import pandas as pd\n+import itertools as it\n+import scipy.stats as st\n+import collections\n+import lxml.etree as ET\n+import pickle as pk\n+import math\n+import os\n+import argparse\n+from svglib.svglib import svg2rlg\n+from reportlab.graphics import renderPDF\n+\n+########################## argparse ##########################################\n+\n+def process_args(args):\n+ parser = argparse.ArgumentParser(usage = \'%(prog)s [options]\',\n+ description = \'process some value\\\'s\'+\n+ \' genes to create a comparison\\\'s map.\')\n+ parser.add_argument(\'-rs\', \'--rules_selector\', \n+ type = str,\n+ default = \'HMRcore\',\n+ choices = [\'HMRcore\', \'Recon\', \'Custom\'], \n+ help = \'chose which type of dataset you want use\')\n+ parser.add_argument(\'-cr\', \'--custom\',\n+ type = str,\n+ help=\'your dataset if you want custom rules\')\n+ parser.add_argument(\'-na\', \'--names\', \n+ type = str,\n+ nargs = \'+\', \n+ help = \'input names\')\n+ parser.add_argument(\'-n\', \'--none\',\n+ type = str,\n+ default = \'true\',\n+ choices = [\'true\', \'false\'], \n+ help = \'compute Nan values\')\n+ parser.add_argument(\'-pv\' ,\'--pValue\', \n+ type = float, \n+ default = 0.05, \n+ help = \'P-Value threshold (default: %(default)s)\')\n+ parser.add_argument(\'-fc\', \'--fChange\', \n+ type = float, \n+ default = 1.5, \n+ help = \'Fold-Change threshold (default: %(default)s)\')\n+ parser.add_argument(\'-td\', \'--tool_dir\',\n+ type = str,\n+ required = True,\n+ help = \'your tool directory\')\n+ parser.add_argument(\'-op\', \'--option\', \n+ type = str, \n+ choices = [\'datasets\', \'dataset_class\', \'datasets_rasonly\'],\n+ help=\'dataset or dataset and class\')\n+ parser.add_argument(\'-ol\', \'--out_log\', \n+ help = "Output log") \n+ parser.add_argument(\'-ids\', \'--input_datas\', \n+ type = str,\n+ nargs = \'+\', \n+ help = \'input datasets\')\n+ parser.add_argument(\'-id\', \'--input_data\',\n+ type = str,\n+ help = \'input dataset\')\n+ parser.add_argument(\'-ic\', \'--input_class\', \n+ type = str, \n+ help = \'sample group specification\')\n+ parser.add_argument(\'-cm\', \'--custom_map\', \n+ type = str, \n+ help = \'custom map\')\n+ parser.add_argument(\'-yn\', \'--yes_no\', \n+ type = str,\n+ choices = [\'yes\', \'no\'],\n+ help = \'if make or not custom map\')\n+ parser.add_argument(\'-gs\', \'--generate_svg\',\n+ type = str,\n+ default = \'true\',\n+ choices = [\'true\', \'false\'], \n+ help = \'generate svg map\')\n+ parser.add_argument(\'-gp\', \'--generate_pdf\',\n+ type = str,\n+ default = \'true\',\n+ choices = [\'true\', \'false\'], \n+ help = \'generate pdf map\')\n+ parser.add_argument(\'-gr\', \'--generate_ras\',\n+ type = str,\n+ default = \'true\',\n+ choices = [\'true\', \'false\'],\n+ help = \'generate reaction activity score\')\n+ parser.add_argument(\'-sr\', \'--single_ras_file\', \n+ type = s'..b'e, name, None)\n+ ids, rules = load_id_rules(recon.get(type_gene))\n+ elif args.rules_selector == \'Custom\':\n+ genes = data_gene(dataset, type_gene, name, gene_in_rule)\n+ \n+ resolve_rules, err = resolve(genes, rules, ids, resolve_none, name)\n+\n+ create_ras(resolve_rules, name, True, rules, ids)\n+ \n+ if err != None and err:\n+ warning(\'Warning: gene\\n\' + str(err) + \'\\nnot found in class \'\n+ + name + \', the expression level for this gene \' +\n+ \'will be considered NaN\\n\')\n+ \n+ print(\'execution succeded\')\n+ return None\n+ \n+ \n+ elif args.option == \'datasets\':\n+ num = 1\n+ for i, j in zip(args.input_datas, args.names):\n+\n+ name = name_dataset(j, num)\n+ dataset = read_dataset(i, name)\n+\n+ dataset.iloc[:, 0] = (dataset.iloc[:, 0]).astype(str)\n+\n+ type_gene = gene_type(dataset.iloc[0, 0], name) \n+ \n+ if args.rules_selector != \'Custom\':\n+ genes = data_gene(dataset, type_gene, name, None)\n+ ids, rules = load_id_rules(recon.get(type_gene))\n+ elif args.rules_selector == \'Custom\':\n+ genes = data_gene(dataset, type_gene, name, gene_in_rule)\n+ \n+ \n+ resolve_rules, err = resolve(genes, rules, ids, resolve_none, name)\n+\n+ if generate_ras:\n+ create_ras(resolve_rules, name, False, rules, ids)\n+ \n+ if err != None and err:\n+ warning(\'Warning: gene\\n\' + str(err) + \'\\nnot found in class \'\n+ + name + \', the expression level for this gene \' +\n+ \'will be considered NaN\\n\')\n+ if resolve_rules != None:\n+ class_pat[name] = list(map(list, zip(*resolve_rules.values())))\n+ num += 1\n+ elif args.option == \'dataset_class\':\n+ name = \'RNAseq\'\n+ dataset = read_dataset(args.input_data, name)\n+ dataset.iloc[:, 0] = (dataset.iloc[:, 0]).astype(str)\n+ type_gene = gene_type(dataset.iloc[0, 0], name)\n+ classes = read_dataset(args.input_class, \'class\')\n+ if not len(classes.columns) == 2:\n+ warning(\'Warning: more than 2 columns in class file. Extra\' +\n+ \'columns have been disregarded\\n\')\n+ classes = classes.astype(str)\n+ if args.rules_selector != \'Custom\':\n+ genes = data_gene(dataset, type_gene, name, None)\n+ ids, rules = load_id_rules(recon.get(type_gene))\n+ elif args.rules_selector == \'Custom\':\n+ genes = data_gene(dataset, type_gene, name, gene_in_rule)\n+ resolve_rules, err = resolve(genes, rules, ids, resolve_none, name)\n+ if err != None and err:\n+ warning(\'Warning: gene\\n\'+str(err)+\'\\nnot found in class \'\n+ + name + \', the expression level for this gene \' +\n+ \'will be considered NaN\\n\')\n+ if resolve_rules != None:\n+ class_pat = split_class(classes, resolve_rules)\n+ if generate_ras:\n+ create_ras(resolve_rules, name, False, rules, ids)\n+ \n+ \t\n+ if args.rules_selector == \'Custom\':\n+ if args.yes_no == \'yes\':\n+ try:\n+ core_map = ET.parse(args.custom_map)\n+ except (ET.XMLSyntaxError, ET.XMLSchemaParseError):\n+ sys.exit(\'Execution aborted: custom map in wrong format\')\n+ elif args.yes_no == \'no\':\n+ core_map = ET.parse(args.tool_dir + \'/local/HMRcoreMap.svg\')\n+ else: \n+ core_map = ET.parse(args.tool_dir+\'/local/HMRcoreMap.svg\')\n+ \n+ maps(core_map, class_pat, ids, args.pValue, args.fChange, create_svg, create_pdf)\n+ \n+ print(\'Execution succeded\')\n+\n+ return None\n+\n+###############################################################################\n+\n+if __name__ == "__main__":\n+ main()\n' |