Source Code for Module bpm.cmdargs.bpms

 1  ''' 
 2  'bpms.py' sets up the command line arguments for the 'genecentric-bpms' 
 3  program. 
 4   
 5  It can do quite a few things in parallel (like generating random bipartitions), 
 6  so this module also does some preprocessing to setup sane defaults for 
 7  parallelization. 
 8  ''' 
 9  import argparse 
10  import multiprocessing as mp 
11   
12  import bpm 
13  from bpm.cmdargs import assert_read_access 
14   
15  try: 
16      __cpus = mp.cpu_count() 
17  except NotImplementedError: 
18      __cpus = 1 
19   
20  parser = argparse.ArgumentParser( 
21      description='BPM generator', 
22      formatter_class=argparse.ArgumentDefaultsHelpFormatter) 
23  aa = parser.add_argument 
24  aa('geneinter', type=str, 
25     metavar='INPUT_GENETIC_INTERACTION_FILE', help='Location of the GI file.') 
26  aa('bpm', type=str, 
27     metavar='OUTPUT_BPM_FILE', help='Where the BPM output will be written.') 
28  aa('-e', '--ignore-list', dest='ignore', type=str, default=None, 
29     metavar='IGNORE_FILE', 
30     help='The location of an ignore gene list file. (One gene per line.) ' 
31          'Any genes in this file will be excluded from the set of genes used ' 
32          'to generate BPMs.') 
33  aa('-c', '--gene-ratio', dest='C', type=float, default=0.90, 
34     metavar='RATIO', help='Gene ratio threshold') 
35  aa('-j', '--jaccard', dest='jaccard', type=float, default=0.66, 
36     metavar='JACCARD_INDEX', help='Jaccard Index threshold') 
37  aa('-m', '--num-bipartitions', dest='M', type=int, default=250, 
38     metavar='NUMBER_BIPARTITIONS', help='Number of bipartitions to generate') 
39  aa('--emap', dest='emap_defaults', action='store_true', 
40     help='If set, EMAP default parameters will be used. This overrides all ' 
41          'other parameters.') 
42  aa('--squaring', dest='squaring', action='store_true', 
43     help='If set, genetic interaction scores will be squared. ' 
44          'Squaring typically speeds convergence.') 
45  aa('--minimum-size', dest='min_size', type=int, default=3, 
46     metavar='MIN_SIZE',  
47     help='Minimum size of BPM. Smaller BPMs are pruned. ' 
48          'Set to 0 to disable.') 
49  aa('--maximum-size', dest='max_size', type=int, default=25, 
50     metavar='MAX_SIZE',  
51     help='Maximum size of BPM. Bigger BPMs are pruned. ' 
52          'Set to 0 to disable.') 
53  aa('-p', '--processes', dest='processes', type=int, default=__cpus, 
54     metavar='PROCESSES', 
55     help='The number of processes to run concurrently. If set to ' 
56          '1, the multiprocessing module will not be used.') 
57  aa('--no-jaccard', dest='pruning', action='store_false', 
58     help='If set, no pruning will occur. Note that --minimum-size and ' 
59          '--maximum-size will still have an effect. Set those to 0 to ' 
60          'disable that pruning.') 
61  aa('--no-progress', dest='progress', action='store_false', 
62     help='If set, the progress bar will not be shown.') 
63  aa('-v', '--verbose', dest='verbose', action='store_true', 
64     help='If set, more output will be shown.') 
65   
66  conf = parser.parse_args() 
67   
68  # BS 
69  if conf.emap_defaults: 
70      conf.C = 0.9 
71      conf.jaccard = 0.66 
72      conf.M = 250 
73      conf.squaring = True 
74      conf.min_size = 3 
75      conf.max_size = 25 
76      conf.pruning = True 
77   
78  # Protect the user from themselves. 
79  # If the provided number of processes is larger than the detected number of 
80  # CPUs, forcefully lower it to the number of CPUs. 
81  if conf.processes > __cpus: 
82      conf.processes = __cpus 
83   
84  # Do some error checking on file inputs... 
85  assert_read_access(conf.geneinter) 
86  if conf.ignore: # ignore list is optional 
87      assert_read_access(conf.ignore) 
88   
89  # Set the global conf variable 
90  bpm.conf = conf 
91