Mercurial > repos > peterjc > tmhmm_and_signalp

comparison tools/protein_analysis/signalp3.py @ 11:99b82a2b1272 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded v0.2.0 which added PSORTb wrapper (written with Konrad Paszkiewicz)
author peterjc
date Wed, 03 Apr 2013 10:49:10 -0400
parents e52220a9ddad
children eb6ac44d4b8e
comparison
equal deleted inserted replaced
10:09ff180d1615 11:99b82a2b1272
61 61
62 FASTA_CHUNK = 500 62 FASTA_CHUNK = 500
63 MAX_LEN = 6000 #Found by trial and error 63 MAX_LEN = 6000 #Found by trial and error
64 64
65 if len(sys.argv) not in [6,8]: 65 if len(sys.argv) not in [6,8]:
66 stop_err("Require five (or 7) arguments, organism, truncate, threads, " 66 stop_err("Require five (or 7) arguments, organism, truncate, threads, "
67 "input protein FASTA file & output tabular file (plus " 67 "input protein FASTA file & output tabular file (plus "
68 "optionally cut method and GFF3 output file). " 68 "optionally cut method and GFF3 output file). "
69 "Got %i arguments." % (len(sys.argv)-1)) 69 "Got %i arguments." % (len(sys.argv)-1))
70 70
71 organism = sys.argv[1] 71 organism = sys.argv[1]
72 if organism not in ["euk", "gram+", "gram-"]: 72 if organism not in ["euk", "gram+", "gram-"]:
73 stop_err("Organism argument %s is not one of euk, gram+ or gram-" % organism) 73 stop_err("Organism argument %s is not one of euk, gram+ or gram-" % organism)
74 74
75 try: 75 try:
76 truncate = int(sys.argv[2]) 76 truncate = int(sys.argv[2])
77 except: 77 except:
78 truncate = 0 78 truncate = 0
79 if truncate < 0: 79 if truncate < 0:
80 stop_err("Truncate argument %s is not a positive integer (or zero)" % sys.argv[2]) 80 stop_err("Truncate argument %s is not a positive integer (or zero)" % sys.argv[2])
81 81
82 num_threads = thread_count(sys.argv[3], default=4) 82 num_threads = thread_count(sys.argv[3], default=4)
83 fasta_file = sys.argv[4] 83 fasta_file = sys.argv[4]
84 tabular_file = sys.argv[5] 84 tabular_file = sys.argv[5]
85 85
86 if len(sys.argv) == 8: 86 if len(sys.argv) == 8:
87 cut_method = sys.argv[6] 87 cut_method = sys.argv[6]
88 if cut_method not in ["NN_Cmax", "NN_Ymax", "NN_Smax", "HMM_Cmax"]: 88 if cut_method not in ["NN_Cmax", "NN_Ymax", "NN_Smax", "HMM_Cmax"]:
89 stop_err("Invalid cut method %r" % cut_method) 89 stop_err("Invalid cut method %r" % cut_method)
90 gff3_file = sys.argv[7] 90 gff3_file = sys.argv[7]
91 else: 91 else:
92 cut_method = None 92 cut_method = None
93 gff3_file = None 93 gff3_file = None
94 94
95 95
96 tmp_dir = tempfile.mkdtemp() 96 tmp_dir = tempfile.mkdtemp()
97 97
98 def clean_tabular(raw_handle, out_handle, gff_handle=None, cut_method=None): 98 def clean_tabular(raw_handle, out_handle, gff_handle=None, cut_method=None):
99 """Clean up SignalP output to make it tabular.""" 99 """Clean up SignalP output to make it tabular."""
100 if cut_method: 100 if cut_method:
101 cut_col = {"NN_Cmax" : 2, 101 cut_col = {"NN_Cmax" : 2,
102 "NN_Ymax" : 5, 102 "NN_Ymax" : 5,
103 "NN_Smax" : 8, 103 "NN_Smax" : 8,
104 "HMM_Cmax" : 16}[cut_method] 104 "HMM_Cmax" : 16}[cut_method]
105 else: 105 else:
106 cut_col = None 106 cut_col = None
107 for line in raw_handle: 107 for line in raw_handle:
108 if not line or line.startswith("#"): 108 if not line or line.startswith("#"):
109 continue 109 continue
110 parts = line.rstrip("\r\n").split() 110 parts = line.rstrip("\r\n").split()
111 assert len(parts)==21, repr(line) 111 assert len(parts)==21, repr(line)
112 assert parts[14].startswith(parts[0]) 112 assert parts[14].startswith(parts[0]), \
113 "Bad entry in SignalP output, ID miss-match:\n%r" % line
113 #Remove redundant truncated name column (col 0) 114 #Remove redundant truncated name column (col 0)
114 #and put full name at start (col 14) 115 #and put full name at start (col 14)
115 parts = parts[14:15] + parts[1:14] + parts[15:] 116 parts = parts[14:15] + parts[1:14] + parts[15:]
116 out_handle.write("\t".join(parts) + "\n") 117 out_handle.write("\t".join(parts) + "\n")
117 118
216 data_handle.close() 217 data_handle.close()
217 out_handle.close() 218 out_handle.close()
218 219
219 #GFF3: 220 #GFF3:
220 if cut_method: 221 if cut_method:
221 make_gff(fasta_file, tabular_file, gff3_file, cut_method) 222 make_gff(fasta_file, tabular_file, gff3_file, cut_method)
222 223
223 clean_up(fasta_files + temp_files) 224 clean_up(fasta_files + temp_files)