Mercurial > repos > peterjc > tmhmm_and_signalp

comparison tools/protein_analysis/seq_analysis_utils.py @ 7:5e62aefb2918 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded v0.1.2 to Test Tool Shed
author peterjc
date Tue, 26 Mar 2013 14:24:56 -0400
parents 39a6e46cdda3
children 41a42022f815
comparison
equal deleted inserted replaced
6:39a6e46cdda3 7:5e62aefb2918
16 16
17 def stop_err(msg, error_level=1): 17 def stop_err(msg, error_level=1):
18 """Print error message to stdout and quit with given error level.""" 18 """Print error message to stdout and quit with given error level."""
19 sys.stderr.write("%s\n" % msg) 19 sys.stderr.write("%s\n" % msg)
20 sys.exit(error_level) 20 sys.exit(error_level)
21
22 try:
23 from multiprocessing import cpu_count
24 except ImportError:
25 #Must be under Python 2.5, this is copied from multiprocessing:
26 def cpu_count():
27 """Returns the number of CPUs in the system."""
28 if sys.platform == 'win32':
29 try:
30 num = int(os.environ['NUMBER_OF_PROCESSORS'])
31 except (ValueError, KeyError):
32 num = 0
33 elif 'bsd' in sys.platform or sys.platform == 'darwin':
34 comm = '/sbin/sysctl -n hw.ncpu'
35 if sys.platform == 'darwin':
36 comm = '/usr' + comm
37 try:
38 with os.popen(comm) as p:
39 num = int(p.read())
40 except ValueError:
41 num = 0
42 else:
43 try:
44 num = os.sysconf('SC_NPROCESSORS_ONLN')
45 except (ValueError, OSError, AttributeError):
46 num = 0
47
48 if num >= 1:
49 return num
50 else:
51 raise NotImplementedError('cannot determine number of cpus')
52
53
54 def thread_count(command_line_arg, default=1):
55 try:
56 num = int(command_line_arg)
57 except:
58 num = default
59 if num < 1:
60 stop_err("Threads argument %r is not a positive integer" % command_line_arg)
61 #Cap this with the pysical limit of the machine,
62 try:
63 num = min(num, cpu_count())
64 except NotImplementedError:
65 pass
66 #For debugging,
67 #hostname = os.environ.get("HOSTNAME", "this machine")
68 #sys.stderr.write("Using %i cores on %s\n" % (num, hostname))
69 return num
70
21 71
22 def fasta_iterator(filename, max_len=None, truncate=None): 72 def fasta_iterator(filename, max_len=None, truncate=None):
23 """Simple FASTA parser yielding tuples of (title, sequence) strings.""" 73 """Simple FASTA parser yielding tuples of (title, sequence) strings."""
24 handle = open(filename) 74 handle = open(filename)
25 title, seq = "", "" 75 title, seq = "", ""
98 pass 148 pass
99 for f in files: 149 for f in files:
100 if os.path.isfile(f): 150 if os.path.isfile(f):
101 os.remove(f) 151 os.remove(f)
102 raise err 152 raise err
153 for f in files:
154 assert os.path.isfile(f), "Missing split file %r (!??)" % f
103 return files 155 return files
104 156
105 def run_jobs(jobs, threads, pause=10, verbose=False): 157 def run_jobs(jobs, threads, pause=10, verbose=False):
106 """Takes list of cmd strings, returns dict with error levels.""" 158 """Takes list of cmd strings, returns dict with error levels."""
107 pending = jobs[:] 159 pending = jobs[:]
108 running = [] 160 running = []
109 results = {} 161 results = {}
162 if threads == 1:
163 #Special case this for speed, don't need the waits
164 for cmd in jobs:
165 results[cmd] = subprocess.call(cmd, shell=True)
166 return results
110 while pending or running: 167 while pending or running:
111 #See if any have finished 168 #See if any have finished
112 for (cmd, process) in running: 169 for (cmd, process) in running:
113 return_code = process.poll() #non-blocking 170 return_code = process.poll() #non-blocking
114 if return_code is not None: 171 if return_code is not None: