Mercurial > repos > peterjc > tmhmm_and_signalp

comparison tools/protein_analysis/promoter2.py @ 21:238eae32483c draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"Check this is up to date with all 2020 changes (black etc)"
author peterjc
date Thu, 17 Jun 2021 08:21:06 +0000
parents a19b3ded8f33
children
comparison
equal deleted inserted replaced
20:a19b3ded8f33 21:238eae32483c
39 39
40 if "-v" in sys.argv or "--version" in sys.argv: 40 if "-v" in sys.argv or "--version" in sys.argv:
41 sys.exit(os.system("promoter -V")) 41 sys.exit(os.system("promoter -V"))
42 42
43 if len(sys.argv) != 4: 43 if len(sys.argv) != 4:
44 sys.exit("Require three arguments, number of threads (int), input DNA FASTA file & output tabular file. " 44 sys.exit(
45 "Got %i arguments." % (len(sys.argv) - 1)) 45 "Require three arguments, number of threads (int), input DNA FASTA "
46 "file & output tabular file. Got %i arguments." % (len(sys.argv) - 1)
47 )
46 48
47 num_threads = thread_count(sys.argv[3], default=4) 49 num_threads = thread_count(sys.argv[3], default=4)
48 fasta_file = os.path.abspath(sys.argv[2]) 50 fasta_file = os.path.abspath(sys.argv[2])
49 tabular_file = os.path.abspath(sys.argv[3]) 51 tabular_file = os.path.abspath(sys.argv[3])
50 52
94 position, score, likelihood = line.strip().split(None, 2) 96 position, score, likelihood = line.strip().split(None, 2)
95 except ValueError: 97 except ValueError:
96 print("WARNING: Problem with line: %r" % line) 98 print("WARNING: Problem with line: %r" % line)
97 continue 99 continue
98 # sys.exit("ERROR: Problem with line: %r" % line) 100 # sys.exit("ERROR: Problem with line: %r" % line)
99 if likelihood not in ["ignored", 101 if likelihood not in [
100 "Marginal prediction", 102 "ignored",
101 "Medium likely prediction", 103 "Marginal prediction",
102 "Highly likely prediction"]: 104 "Medium likely prediction",
105 "Highly likely prediction",
106 ]:
103 sys.exit("ERROR: Problem with line: %r" % line) 107 sys.exit("ERROR: Problem with line: %r" % line)
104 out_handle.write("%s\t%s\t%s\t%s\n" % (identifier, position, score, likelihood)) 108 out_handle.write(
109 "%s\t%s\t%s\t%s\n" % (identifier, position, score, likelihood)
110 )
105 return queries 111 return queries
106 112
107 113
108 working_dir, bin = get_path_and_binary() 114 working_dir, bin = get_path_and_binary()
109 115
112 118
113 # Note that if the input FASTA file contains no sequences, 119 # Note that if the input FASTA file contains no sequences,
114 # split_fasta returns an empty list (i.e. zero temp files). 120 # split_fasta returns an empty list (i.e. zero temp files).
115 # We deliberately omit the FASTA descriptions to avoid a 121 # We deliberately omit the FASTA descriptions to avoid a
116 # bug in promoter2 with descriptions over 200 characters. 122 # bug in promoter2 with descriptions over 200 characters.
117 fasta_files = split_fasta(fasta_file, os.path.join(tmp_dir, "promoter"), FASTA_CHUNK, keep_descr=False) 123 fasta_files = split_fasta(
124 fasta_file, os.path.join(tmp_dir, "promoter"), FASTA_CHUNK, keep_descr=False
125 )
118 temp_files = [f + ".out" for f in fasta_files] 126 temp_files = [f + ".out" for f in fasta_files]
119 jobs = ["%s %s > %s" % (bin, fasta, temp) 127 jobs = [
120 for fasta, temp in zip(fasta_files, temp_files)] 128 "%s %s > %s" % (bin, fasta, temp) for fasta, temp in zip(fasta_files, temp_files)
129 ]
121 130
122 131
123 def clean_up(file_list): 132 def clean_up(file_list):
124 for f in file_list: 133 for f in file_list:
125 if os.path.isfile(f): 134 if os.path.isfile(f):
143 try: 152 try:
144 output = open(temp).readline() 153 output = open(temp).readline()
145 except IOError: 154 except IOError:
146 output = "" 155 output = ""
147 clean_up(fasta_files + temp_files) 156 clean_up(fasta_files + temp_files)
148 sys.exit("One or more tasks failed, e.g. %i from %r gave:\n%s" % (error_level, cmd, output), 157 sys.exit(
149 error_level) 158 "One or more tasks failed, e.g. %i from %r gave:\n%s"
159 % (error_level, cmd, output),
160 error_level,
161 )
150 162
151 del results 163 del results
152 del jobs 164 del jobs
153 165
154 out_handle = open(tabular_file, "w") 166 out_handle = open(tabular_file, "w")