[2] | 1 | #!/usr/bin/env python
|
---|
| 2 | #Dan Blankenberg
|
---|
| 3 |
|
---|
| 4 | import sys, os
|
---|
| 5 |
|
---|
| 6 | assert sys.version_info[:2] >= ( 2, 4 )
|
---|
| 7 |
|
---|
| 8 | def __main__():
|
---|
| 9 | base_dir = os.path.join( os.getcwd(), "bacteria" )
|
---|
| 10 | try:
|
---|
| 11 | base_dir = sys.argv[1]
|
---|
| 12 | except:
|
---|
| 13 | pass
|
---|
| 14 | #print "using default base_dir:", base_dir
|
---|
| 15 |
|
---|
| 16 | organisms = {}
|
---|
| 17 | for result in os.walk(base_dir):
|
---|
| 18 | this_base_dir,sub_dirs,files = result
|
---|
| 19 | for file in files:
|
---|
| 20 | if file[-5:] == ".info":
|
---|
| 21 | dict = {}
|
---|
| 22 | info_file = open(os.path.join(this_base_dir,file),'r')
|
---|
| 23 | info = info_file.readlines()
|
---|
| 24 | info_file.close()
|
---|
| 25 | for line in info:
|
---|
| 26 | fields = line.replace("\n","").split("=")
|
---|
| 27 | dict[fields[0]]="=".join(fields[1:])
|
---|
| 28 | if 'genome project id' in dict.keys():
|
---|
| 29 | name = dict['genome project id']
|
---|
| 30 | if 'build' in dict.keys():
|
---|
| 31 | name = dict['build']
|
---|
| 32 | if name not in organisms.keys():
|
---|
| 33 | organisms[name] = {'chrs':{},'base_dir':this_base_dir}
|
---|
| 34 | for key in dict.keys():
|
---|
| 35 | organisms[name][key]=dict[key]
|
---|
| 36 | else:
|
---|
| 37 | if dict['organism'] not in organisms.keys():
|
---|
| 38 | organisms[dict['organism']] = {'chrs':{},'base_dir':this_base_dir}
|
---|
| 39 | organisms[dict['organism']]['chrs'][dict['chromosome']]=dict
|
---|
| 40 |
|
---|
| 41 | orgs = organisms.keys()
|
---|
| 42 | for org in orgs:
|
---|
| 43 | if 'name' not in organisms[org]:del organisms[org]
|
---|
| 44 |
|
---|
| 45 | orgs = organisms.keys()
|
---|
| 46 | #need to sort by name
|
---|
| 47 | swap_test = False
|
---|
| 48 | for i in range(0, len(orgs) - 1):
|
---|
| 49 | for j in range(0, len(orgs) - i - 1):
|
---|
| 50 | if organisms[orgs[j]]['name'] > organisms[orgs[j + 1]]['name']:
|
---|
| 51 | orgs[j], orgs[j + 1] = orgs[j + 1], orgs[j]
|
---|
| 52 | swap_test = True
|
---|
| 53 | if swap_test == False:
|
---|
| 54 | break
|
---|
| 55 |
|
---|
| 56 |
|
---|
| 57 |
|
---|
| 58 |
|
---|
| 59 | print "||'''Organism'''||'''Kingdom'''||'''Group'''||'''Links to UCSC Archaea Browser'''||"
|
---|
| 60 |
|
---|
| 61 |
|
---|
| 62 | for org in orgs:
|
---|
| 63 | org = organisms[org]
|
---|
| 64 | at_ucsc = False
|
---|
| 65 | #if no gpi, then must be a ncbi chr which corresponds to a UCSC org, w/o matching UCSC designation
|
---|
| 66 | try:
|
---|
| 67 | build = org['genome project id']
|
---|
| 68 | except: continue
|
---|
| 69 | if 'build' in org:
|
---|
| 70 | build = org['build']
|
---|
| 71 | at_ucsc = True
|
---|
| 72 |
|
---|
| 73 | out_str = "||"+org['name']+"||"+org['kingdom']+"||"+org['group']+"||"
|
---|
| 74 | if at_ucsc:
|
---|
| 75 | out_str = out_str + "Yes"
|
---|
| 76 | out_str = out_str + "||"
|
---|
| 77 | print out_str
|
---|
| 78 |
|
---|
| 79 | if __name__ == "__main__": __main__()
|
---|