effectivet3: tools/effectiveT3/effectiveT3.py annotate

annotate tools/effectiveT3/effectiveT3.py @ 11:ed8c1babc166 draft default tip

v0.0.21 - Added bio.tools xref

author	peterjc
date	Tue, 12 Mar 2024 16:04:13 +0000
parents	a46d7861c32c
children

rev	line source
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	1 #!/usr/bin/env python
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	2 """Wrapper for EffectiveT3 v1.0.1 for use in Galaxy.
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	3
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	4 This script takes exactly five command line arguments:
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	5 * model name (e.g. TTSS_STD-1.0.1.jar)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	6 * threshold (selective or sensitive)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	7 * an input protein FASTA filename
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	8 * output tabular filename
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	9
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	10 It then calls the standalone Effective T3 v1.0.1 program (not the
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	11 webservice), and reformats the semi-colon separated output into
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	12 tab separated output for use in Galaxy.
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	13 """
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	14 import os
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	15
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	16 # We want to be able to use shutil.which, but need Python 3.3+
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	17 # import shutil
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	18 import subprocess
9 512530020360 v0.0.18 Internal changes to command line handling peterjc parents: 8 diff changeset	19 import sys
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	20
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	21 # The Galaxy auto-install via tool_dependencies.xml will set the
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	22 # environment variable $EFFECTIVET3 pointing at the folder with
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	23 # the JAR file.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	24 #
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	25 # The BioConda recipe will put a wrapper script on the $PATH,
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	26 # which we can use to find the JAR file.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	27 #
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	28 # We fall back on /opt/EffectiveT3/
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	29 #
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	30 effective_t3_jarname = "TTSS_GUI-1.0.1.jar"
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	31
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	32 if "-v" in sys.argv or "--version" in sys.argv:
5 1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	33 # TODO - Get version of the JAR file dynamically?
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	34 print("Wrapper v0.0.20, for %s" % effective_t3_jarname)
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	35 sys.exit(0)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	36
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	37 if len(sys.argv) != 5:
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	38 sys.exit(
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	39 "Require four arguments: model, threshold, input protein "
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	40 "FASTA file & output tabular file"
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	41 )
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	42
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	43 model, threshold, fasta_file, tabular_file = sys.argv[1:]
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	44
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	45 if not os.path.isfile(fasta_file):
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	46 sys.exit("Input FASTA file not found: %s" % fasta_file)
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	47
9 512530020360 v0.0.18 Internal changes to command line handling peterjc parents: 8 diff changeset	48 if threshold not in ["selective", "sensitive"] and not threshold.startswith("cutoff="):
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	49 sys.exit(
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	50 "Threshold should be selective, sensitive, or cutoff=..., not %r" % threshold
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	51 )
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	52
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	53
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	54 def clean_tabular(raw_handle, out_handle):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	55 """Clean up Effective T3 output to make it tabular."""
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	56 count = 0
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	57 positive = 0
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	58 errors = 0
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	59 for line in raw_handle:
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	60 if (
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	61 not line
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	62 or line.startswith("#")
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	63 or line.startswith("Id; Description; Score;")
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	64 ):
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	65 continue
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	66 assert line.count(";") >= 3, repr(line)
5 1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	67 # Normally there will just be three semi-colons, however the
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	68 # original FASTA file's ID or description might have had
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	69 # semi-colons in it as well, hence the following hackery:
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	70 try:
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	71 id_descr, score, effective = line.rstrip("\r\n").rsplit(";", 2)
5 1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	72 # Cope when there was no FASTA description
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	73 if "; " not in id_descr and id_descr.endswith(";"):
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	74 id = id_descr[:-1]
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	75 descr = ""
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	76 else:
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	77 id, descr = id_descr.split("; ", 1)
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	78 except ValueError:
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	79 sys.exit("Problem parsing line:\n%s\n" % line)
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	80 parts = [s.strip() for s in [id, descr, score, effective]]
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	81 out_handle.write("\t".join(parts) + "\n")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	82 count += 1
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	83 if float(score) < 0:
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	84 errors += 1
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	85 if effective.lower() == "true":
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	86 positive += 1
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	87 return count, positive, errors
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	88
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	89
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	90 def run(cmd):
9 512530020360 v0.0.18 Internal changes to command line handling peterjc parents: 8 diff changeset	91 """Run the command line string via subprocess."""
5 1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	92 # Avoid using shell=True when we call subprocess to ensure if the Python
1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	93 # script is killed, so too is the child process.
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	94 try:
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	95 child = subprocess.Popen(
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	96 cmd, universal_newlines=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	97 )
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	98 except Exception as err:
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	99 sys.exit("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
5 1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	100 # Use .communicate as can get deadlocks with .wait(),
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	101 stdout, stderr = child.communicate()
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	102 return_code = child.returncode
7 5f85301d50bf v0.0.16, adding new model TTSS-STD-2.0.2.jar peterjc parents: 6 diff changeset	103 if return_code or stderr.startswith("Exception in thread"):
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	104 cmd_str = " ".join(cmd) # doesn't quote spaces etc
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	105 if stderr and stdout:
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	106 sys.exit(
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	107 "Return code %i from command:\n%s\n\n%s\n\n%s"
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	108 % (return_code, cmd_str, stdout, stderr)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	109 )
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	110 else:
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	111 sys.exit(
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	112 "Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	113 )
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	114
7 5f85301d50bf v0.0.16, adding new model TTSS-STD-2.0.2.jar peterjc parents: 6 diff changeset	115
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	116 try:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	117 from shutil import which
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	118 except ImportError:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	119 # Likely running on Python 2, use backport:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	120 def which(cmd, mode=os.F_OK \| os.X_OK, path=None):
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	121 """Python implementation of command line tool which.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	122
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	123 Given a command, mode, and a PATH string, return the path which
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	124 conforms to the given mode on the PATH, or None if there is no such
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	125 file.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	126
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	127 `mode` defaults to os.F_OK \| os.X_OK. `path` defaults to the result
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	128 of os.environ.get("PATH"), or can be overridden with a custom search
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	129 path.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	130 """
11 ed8c1babc166 v0.0.21 - Added bio.tools xref peterjc parents: 10 diff changeset	131
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	132 # Check that a given file can be accessed with the correct mode.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	133 # Additionally check that `file` is not a directory, as on Windows
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	134 # directories pass the os.access check.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	135 def _access_check(fn, mode):
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	136 return os.path.exists(fn) and os.access(fn, mode) and not os.path.isdir(fn)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	137
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	138 # Short circuit. If we're given a full path which matches the mode
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	139 # and it exists, we're done here.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	140 if _access_check(cmd, mode):
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	141 return cmd
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	142
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	143 path = (path or os.environ.get("PATH", os.defpath)).split(os.pathsep)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	144
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	145 if sys.platform == "win32":
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	146 # The current directory takes precedence on Windows.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	147 if os.curdir not in path:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	148 path.insert(0, os.curdir)
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	149
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	150 # PATHEXT is necessary to check on Windows.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	151 pathext = os.environ.get("PATHEXT", "").split(os.pathsep)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	152 # See if the given file matches any of the expected path extensions.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	153 # This will allow us to short circuit when given "python.exe".
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	154 matches = [cmd for ext in pathext if cmd.lower().endswith(ext.lower())]
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	155 # If it does match, only test that one, otherwise we have to try
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	156 # others.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	157 files = [cmd] if matches else [cmd + ext.lower() for ext in pathext]
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	158 else:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	159 # On other platforms you don't have things like PATHEXT to tell you
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	160 # what file suffixes are executable, so just pass on cmd as-is.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	161 files = [cmd]
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	162
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	163 seen = set()
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	164 for dir in path:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	165 dir = os.path.normcase(dir)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	166 if dir not in seen:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	167 seen.add(dir)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	168 for thefile in files:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	169 name = os.path.join(dir, thefile)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	170 if _access_check(name, mode):
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	171 return name
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	172 return None
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	173
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	174
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	175 # Try in order the following to find the JAR file:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	176 # - Location of any wrapper script, e.g. from BioConda installation
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	177 # - The $EFFECTIVET3 env var, e.g. old-style Galaxy tool installation
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	178 # - The /opt/EffectiveT3/ folder.
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	179 effective_t3_jar = None
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	180 effective_t3_dir = None
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	181 dirs = ["/opt/EffectiveT3/"]
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	182 if "EFFECTIVET3" in os.environ:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	183 dirs.insert(0, os.environ.get("EFFECTIVET3"))
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	184 if which("effectivet3"):
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	185 # Assuming this is a BioConda installed wrapper for effective T3,
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	186 # this will get the directory of the wrapper script which is where
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	187 # the JAR file will be:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	188 dirs.insert(0, os.path.split(os.path.realpath(which("effectivet3")))[0])
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	189 for effective_t3_dir in dirs:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	190 effective_t3_jar = os.path.join(effective_t3_dir, effective_t3_jarname)
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	191 if os.path.isfile(effective_t3_jar):
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	192 # Good
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	193 break
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	194 effective_t3_jar = None
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	195 if not effective_t3_dir or not effective_t3_jar:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	196 sys.exit("Effective T3 JAR file %r not found in %r" % (effective_t3_jarname, dirs))
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	197
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	198 if not os.path.isdir(os.path.join(effective_t3_dir, "module")):
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	199 sys.exit(
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	200 "Effective T3 module folder not found: %r"
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	201 % os.path.join(effective_t3_dir, "module")
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	202 )
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	203
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	204 effective_t3_model = os.path.join(effective_t3_dir, "module", model)
60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	205 if not os.path.isfile(effective_t3_model):
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	206 sys.stderr.write(
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	207 "Contents of %r is %s\n"
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	208 % (
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	209 os.path.join(effective_t3_dir, "module"),
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	210 ", ".join(
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	211 repr(p) for p in os.listdir(os.path.join(effective_t3_dir, "module"))
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	212 ),
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	213 )
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	214 )
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	215 sys.stderr.write("Main JAR was found: %r\n" % effective_t3_jar)
60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	216 sys.exit("Effective T3 model JAR file not found: %r" % effective_t3_model)
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	217
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	218 # We will have write access wherever the output should be,
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	219 if tabular_file == "/dev/stdout":
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	220 temp_file = os.path.abspath("effectivet3_tabular_output.tmp")
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	221 else:
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	222 temp_file = os.path.abspath(tabular_file + ".tmp")
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	223
5 1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	224 # Use absolute paths since will change current directory...
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	225 tabular_file = os.path.abspath(tabular_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	226 fasta_file = os.path.abspath(fasta_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	227
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	228 cmd = [
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	229 "java",
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	230 "-jar",
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	231 effective_t3_jar,
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	232 "-f",
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	233 fasta_file,
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	234 "-m",
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	235 model,
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	236 "-t",
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	237 threshold,
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	238 "-o",
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	239 temp_file,
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	240 "-q",
a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	241 ]
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	242
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	243 try:
5 1ea715da1879 Uploaded v0.0.13, embed citation, relax test for floating point differences peterjc parents: 3 diff changeset	244 # Must run from directory above the module subfolder:
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	245 os.chdir(effective_t3_dir)
60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	246 except Exception:
60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	247 sys.exit("Could not change to Effective T3 folder: %s" % effective_t3_dir)
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	248
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	249 run(cmd)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	250
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	251 if not os.path.isfile(temp_file):
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	252 sys.exit("ERROR - No output file from Effective T3")
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	253
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	254 out_handle = open(tabular_file, "w")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	255 out_handle.write("#ID\tDescription\tScore\tEffective\n")
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	256 data_handle = open(temp_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	257 count, positive, errors = clean_tabular(data_handle, out_handle)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	258 data_handle.close()
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	259 out_handle.close()
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	260
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	261 os.remove(temp_file)
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	262
b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	263 if errors:
10 a46d7861c32c "Update all the pico_galaxy tools on main Tool Shed" peterjc parents: 9 diff changeset	264 print("%i sequences, %i positive, %i errors" % (count, positive, errors))
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	265 else:
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	266 print("%i/%i sequences positive" % (positive, count))
3 b0b927299aee Uploaded v0.0.11 with automatic dependency installation. peterjc parents: diff changeset	267
8 60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	268 if count and count == errors:
60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	269 # Galaxy will still allow them to see the output file
60a9b3f760cc v0.0.17 Used cached URL, python style updates peterjc parents: 7 diff changeset	270 sys.exit("All your sequences gave an error code")

Mercurial > repos > peterjc > effectivet3

annotate tools/effectiveT3/effectiveT3.py @ 11:ed8c1babc166 draft default tip