tmhmm_and_signalp: tools/protein_analysis/seq_analysis

annotate tools/protein_analysis/seq_analysis_utils.py @ 27:9e36a1b9302d draft

planemo upload for repository https://github.com/peterjc/pico_galaxy/tools/protein_analysis commit 3c6f0dca0e1318eecd1e07d177ffc5752b4f6c95

author	peterjc
date	Thu, 21 May 2015 10:57:40 -0400
parents	20139cb4c844
children	3cb02adf4326

rev	line source
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	1 """A few useful functions for working with FASTA files and running jobs.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	2
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	3 This module was originally written to hold common code used in both the TMHMM
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	4 and SignalP wrappers in Galaxy.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	5
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	6 Given Galaxy currently supports Python 2.4+ this cannot use the Python module
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	7 multiprocessing so the function run_jobs instead is a simple pool approach
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	8 using just the subprocess library.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	9 """
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	10 import sys
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	11 import os
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	12 import subprocess
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	13 from time import sleep
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	14
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	15 __version__ = "0.0.1"
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	16
26 20139cb4c844 planemo upload for repository https://github.com/peterjc/pico_galaxy/tools/protein_analysis commit 221d4187992cbb993e02dc3ea0ef0150c7916a4a-dirty peterjc parents: 25 diff changeset	17 def sys_exit(msg, error_level=1):
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	18 """Print error message to stdout and quit with given error level."""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	19 sys.stderr.write("%s\n" % msg)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	20 sys.exit(error_level)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	21
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	22 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	23 from multiprocessing import cpu_count
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	24 except ImportError:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	25 #Must be under Python 2.5, this is copied from multiprocessing:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	26 def cpu_count():
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	27 """Returns the number of CPUs in the system."""
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	28 if sys.platform == 'win32':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	29 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	30 num = int(os.environ['NUMBER_OF_PROCESSORS'])
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	31 except (ValueError, KeyError):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	32 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	33 elif 'bsd' in sys.platform or sys.platform == 'darwin':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	34 comm = '/sbin/sysctl -n hw.ncpu'
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	35 if sys.platform == 'darwin':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	36 comm = '/usr' + comm
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	37 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	38 with os.popen(comm) as p:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	39 num = int(p.read())
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	40 except ValueError:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	41 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	42 else:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	43 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	44 num = os.sysconf('SC_NPROCESSORS_ONLN')
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	45 except (ValueError, OSError, AttributeError):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	46 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	47
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	48 if num >= 1:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	49 return num
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	50 else:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	51 raise NotImplementedError('cannot determine number of cpus')
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	52
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	53
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	54 def thread_count(command_line_arg, default=1):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	55 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	56 num = int(command_line_arg)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	57 except:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	58 num = default
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	59 if num < 1:
26 20139cb4c844 planemo upload for repository https://github.com/peterjc/pico_galaxy/tools/protein_analysis commit 221d4187992cbb993e02dc3ea0ef0150c7916a4a-dirty peterjc parents: 25 diff changeset	60 sys_exit("Threads argument %r is not a positive integer" % command_line_arg)
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	61 #Cap this with the pysical limit of the machine,
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	62 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	63 num = min(num, cpu_count())
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	64 except NotImplementedError:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	65 pass
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	66 #For debugging,
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	67 #hostname = os.environ.get("HOSTNAME", "this machine")
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	68 #sys.stderr.write("Using %i cores on %s\n" % (num, hostname))
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	69 return num
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	70
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	71
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	72 def fasta_iterator(filename, max_len=None, truncate=None):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	73 """Simple FASTA parser yielding tuples of (title, sequence) strings."""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	74 handle = open(filename)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	75 title, seq = "", ""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	76 for line in handle:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	77 if line.startswith(">"):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	78 if title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	79 if truncate:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	80 seq = seq[:truncate]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	81 if max_len and len(seq) > max_len:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	82 raise ValueError("Sequence %s is length %i, max length %i" \
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	83 % (title.split()[0], len(seq), max_len))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	84 yield title, seq
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	85 title = line[1:].rstrip()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	86 seq = ""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	87 elif title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	88 seq += line.strip()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	89 elif not line.strip() or line.startswith("#"):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	90 #Ignore blank lines, and any comment lines
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	91 #between records (starting with hash).
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	92 pass
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	93 else:
25 41a42022f815 Uploaded v0.2.6, embedded citations peterjc parents: 7 diff changeset	94 handle.close()
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	95 raise ValueError("Bad FASTA line %r" % line)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	96 handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	97 if title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	98 if truncate:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	99 seq = seq[:truncate]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	100 if max_len and len(seq) > max_len:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	101 raise ValueError("Sequence %s is length %i, max length %i" \
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	102 % (title.split()[0], len(seq), max_len))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	103 yield title, seq
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	104 raise StopIteration
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	105
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	106 def split_fasta(input_filename, output_filename_base, n=500, truncate=None, keep_descr=False, max_len=None):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	107 """Split FASTA file into sub-files each of at most n sequences.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	108
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	109 Returns a list of the filenames used (based on the input filename).
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	110 Each sequence can also be truncated (since we only need the start for
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	111 SignalP), and have its description discarded (since we don't usually
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	112 care about it and some tools don't like very long title lines).
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	113
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	114 If a max_len is given and any sequence exceeds it no temp files are
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	115 created and an exception is raised.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	116 """
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	117 iterator = fasta_iterator(input_filename, max_len, truncate)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	118 files = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	119 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	120 while True:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	121 records = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	122 for i in range(n):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	123 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	124 records.append(iterator.next())
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	125 except StopIteration:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	126 break
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	127 if not records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	128 break
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	129 new_filename = "%s.%i.tmp" % (output_filename_base, len(files))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	130 handle = open(new_filename, "w")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	131 if keep_descr:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	132 for title, seq in records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	133 handle.write(">%s\n" % title)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	134 for i in range(0, len(seq), 60):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	135 handle.write(seq[i:i+60] + "\n")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	136 else:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	137 for title, seq in records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	138 handle.write(">%s\n" % title.split()[0])
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	139 for i in range(0, len(seq), 60):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	140 handle.write(seq[i:i+60] + "\n")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	141 handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	142 files.append(new_filename)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	143 #print "%i records in %s" % (len(records), new_filename)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	144 except ValueError, err:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	145 #Max length failure from parser - clean up
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	146 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	147 handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	148 except:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	149 pass
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	150 for f in files:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	151 if os.path.isfile(f):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	152 os.remove(f)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	153 raise err
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	154 for f in files:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	155 assert os.path.isfile(f), "Missing split file %r (!??)" % f
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	156 return files
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	157
6 39a6e46cdda3 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: 3 diff changeset	158 def run_jobs(jobs, threads, pause=10, verbose=False):
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	159 """Takes list of cmd strings, returns dict with error levels."""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	160 pending = jobs[:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	161 running = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	162 results = {}
7 5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	163 if threads == 1:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	164 #Special case this for speed, don't need the waits
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	165 for cmd in jobs:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	166 results[cmd] = subprocess.call(cmd, shell=True)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed peterjc parents: 6 diff changeset	167 return results
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	168 while pending or running:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	169 #See if any have finished
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	170 for (cmd, process) in running:
3 fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	171 return_code = process.poll() #non-blocking
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	172 if return_code is not None:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	173 results[cmd] = return_code
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	174 running = [(cmd, process) for (cmd, process) in running \
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	175 if cmd not in results]
3 fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	176 if verbose:
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	177 print "%i jobs pending, %i running, %i completed" \
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	178 % (len(pending), len(running), len(results))
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	179 #See if we can start any new threads
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	180 while pending and len(running) < threads:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	181 cmd = pending.pop(0)
3 fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	182 if verbose:
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	183 print cmd
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	184 process = subprocess.Popen(cmd, shell=True)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	185 running.append((cmd, process))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	186 #Loop...
6 39a6e46cdda3 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: 3 diff changeset	187 sleep(pause)
3 fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	188 if verbose:
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository peterjc parents: 0 diff changeset	189 print "%i jobs completed" % len(results)
0 a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	190 assert set(jobs) == set(results)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	191 return results

Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/seq_analysis_utils.py @ 27:9e36a1b9302d draft