Mercurial > repos > peterjc > tmhmm_and_signalp

comparison tools/protein_analysis/wolf_psort.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 e1996f0f4e85
comparison
equal deleted inserted replaced
20:a19b3ded8f33 21:238eae32483c
67 67
68 if "-v" in sys.argv or "--version" in sys.argv: 68 if "-v" in sys.argv or "--version" in sys.argv:
69 sys.exit("WoLF-PSORT wrapper version 0.0.11") 69 sys.exit("WoLF-PSORT wrapper version 0.0.11")
70 70
71 if len(sys.argv) != 5: 71 if len(sys.argv) != 5:
72 sys.exit("Require four arguments, organism, threads, input protein FASTA file & output tabular file") 72 sys.exit(
73 "Require four arguments, organism, threads, input protein FASTA file, "
74 "and output tabular file"
75 )
73 76
74 organism = sys.argv[1] 77 organism = sys.argv[1]
75 if organism not in ["animal", "plant", "fungi"]: 78 if organism not in ["animal", "plant", "fungi"]:
76 sys.exit("Organism argument %s is not one of animal, plant, fungi" % organism) 79 sys.exit("Organism argument %s is not one of animal, plant, fungi" % organism)
77 80
86 if not line or line.startswith("#"): 89 if not line or line.startswith("#"):
87 continue 90 continue
88 name, data = line.rstrip("\r\n").split(None, 1) 91 name, data = line.rstrip("\r\n").split(None, 1)
89 for rank, comp_data in enumerate(data.split(",")): 92 for rank, comp_data in enumerate(data.split(",")):
90 comp, score = comp_data.split() 93 comp, score = comp_data.split()
91 out_handle.write("%s\t%s\t%s\t%i\n" 94 out_handle.write("%s\t%s\t%s\t%i\n" % (name, comp, score, rank + 1))
92 % (name, comp, score, rank + 1))
93 95
94 96
95 fasta_files = split_fasta(fasta_file, tabular_file, n=FASTA_CHUNK) 97 fasta_files = split_fasta(fasta_file, tabular_file, n=FASTA_CHUNK)
96 temp_files = [f + ".out" for f in fasta_files] 98 temp_files = [f + ".out" for f in fasta_files]
97 assert len(fasta_files) == len(temp_files) 99 assert len(fasta_files) == len(temp_files)
98 jobs = ["%s %s < %s > %s" % (exe, organism, fasta, temp) 100 jobs = [
99 for (fasta, temp) in zip(fasta_files, temp_files)] 101 "%s %s < %s > %s" % (exe, organism, fasta, temp)
102 for (fasta, temp) in zip(fasta_files, temp_files)
103 ]
100 assert len(fasta_files) == len(temp_files) == len(jobs) 104 assert len(fasta_files) == len(temp_files) == len(jobs)
101 105
102 106
103 def clean_up(file_list): 107 def clean_up(file_list):
104 for f in file_list: 108 for f in file_list:
118 except IOError: 122 except IOError:
119 output = "" 123 output = ""
120 if error_level or output.lower().startswith("error running"): 124 if error_level or output.lower().startswith("error running"):
121 clean_up(fasta_files) 125 clean_up(fasta_files)
122 clean_up(temp_files) 126 clean_up(temp_files)
123 sys.exit("One or more tasks failed, e.g. %i from %r gave:\n%s" % (error_level, cmd, output), 127 sys.exit(
124 error_level) 128 "One or more tasks failed, e.g. %i from %r gave:\n%s"
129 % (error_level, cmd, output),
130 error_level,
131 )
125 del results 132 del results
126 133
127 out_handle = open(tabular_file, "w") 134 out_handle = open(tabular_file, "w")
128 out_handle.write("#ID\tCompartment\tScore\tRank\n") 135 out_handle.write("#ID\tCompartment\tScore\tRank\n")
129 for temp in temp_files: 136 for temp in temp_files: