tmhmm_and_signalp: tools/protein_analysis/rxlr

annotate tools/protein_analysis/rxlr_motifs.py @ 16:7de64c8b258d draft

Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub

author	peterjc
date	Wed, 18 Sep 2013 06:16:58 -0400
parents	a290c6d4e658
children	eb6ac44d4b8e

rev	line source
6 a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	1 #!/usr/bin/env python
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	2 """Implements assorted RXLR motif methods from the literature
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	3
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	4 This script takes exactly four command line arguments:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	5 * Protein FASTA filename
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	6 * Number of threads
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	7 * Model name (Bhattacharjee2006, Win2007, Whisson2007)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	8 * Output tabular filename
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	9
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	10 The model names are:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	11
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	12 Bhattacharjee2006: Simple regular expression search for RXLR
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	13 with additional requirements for positioning and signal peptide.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	14
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	15 Win2007: Simple regular expression search for RXLR, but with
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	16 different positional requirements.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	17
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	18 Whisson2007: As Bhattacharjee2006 but with a more complex regular
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	19 expression to look for RXLR-EER domain, and additionally calls HMMER.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	20
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	21 See the help text in the accompanying Galaxy tool XML file for more
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	22 details including the full references.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	23
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	24 Note:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	25
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	26 Bhattacharjee et al. (2006) and Win et al. (2007) used SignalP v2.0,
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	27 which is no longer available. The current release is SignalP v3.0
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	28 (Mar 5, 2007). We have therefore opted to use the NN Ymax position for
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	29 the predicted cleavage site, as this is expected to be more accurate.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	30 Also note that the HMM score values have changed from v2.0 to v3.0.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	31 Whisson et al. (2007) used SignalP v3.0 anyway.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	32
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	33 Whisson et al. (2007) used HMMER 2.3.2, and althought their HMM model
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	34 can still be used with hmmsearch from HMMER 3 this this does give
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	35 slightly different results. We expect the hmmsearch from HMMER 2.3.2
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	36 (the last stable release of HMMER 2) to be present on the path under
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	37 the name hmmsearch2 (allowing it to co-exist with HMMER 3).
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	38 """
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	39 import os
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	40 import sys
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	41 import re
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	42 import subprocess
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	43 from seq_analysis_utils import stop_err, fasta_iterator
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	44
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	45 if len(sys.argv) != 5:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	46 stop_err("Requires four arguments: protein FASTA filename, threads, model, and output filename")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	47
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	48 fasta_file, threads, model, tabular_file = sys.argv[1:]
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	49 hmm_output_file = tabular_file + ".hmm.tmp"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	50 signalp_input_file = tabular_file + ".fasta.tmp"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	51 signalp_output_file = tabular_file + ".tabular.tmp"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	52 min_signalp_hmm = 0.9
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	53 hmmer_search = "hmmsearch2"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	54
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	55 if model == "Bhattacharjee2006":
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	56 signalp_trunc = 70
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	57 re_rxlr = re.compile("R.LR")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	58 min_sp = 10
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	59 max_sp = 40
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	60 max_sp_rxlr = 100
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	61 min_rxlr_start = 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	62 #Allow signal peptide to be at most 40aa, and want RXLR to be
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	63 #within 100aa, therefore for the prescreen the max start is 140:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	64 max_rxlr_start = max_sp + max_sp_rxlr
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	65 elif model == "Win2007":
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	66 signalp_trunc = 70
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	67 re_rxlr = re.compile("R.LR")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	68 min_sp = 10
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	69 max_sp = 40
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	70 min_rxlr_start = 30
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	71 max_rxlr_start = 60
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	72 #No explicit limit on separation of signal peptide clevage
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	73 #and RXLR, but shortest signal peptide is 10, and furthest
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	74 #away RXLR is 60, so effectively limit is 50.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	75 max_sp_rxlr = max_rxlr_start - min_sp + 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	76 elif model == "Whisson2007":
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	77 signalp_trunc = 0 #zero for no truncation
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	78 re_rxlr = re.compile("R.LR.{,40}[ED][ED][KR]")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	79 min_sp = 10
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	80 max_sp = 40
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	81 max_sp_rxlr = 100
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	82 min_rxlr_start = 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	83 max_rxlr_start = max_sp + max_sp_rxlr
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	84 else:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	85 stop_err("Did not recognise the model name %r\n"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	86 "Use Bhattacharjee2006, Win2007, or Whisson2007" % model)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	87
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	88
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	89 def get_hmmer_version(exe, required=None):
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	90 cmd = "%s -h" % exe
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	91 try:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	92 child = subprocess.Popen([exe, "-h"], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	93 except OSError:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	94 raise ValueError("Could not run %s" % exe)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	95 stdout, stderr = child.communicate()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	96 if required:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	97 return required in stdout
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	98 elif "HMMER 2" in stdout:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	99 return 2
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	100 elif "HMMER 3" in stdout:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	101 return 3
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	102 else:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	103 raise ValueError("Could not determine version of %s" % exe)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	104
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	105
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	106 #Run hmmsearch for Whisson et al. (2007)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	107 if model == "Whisson2007":
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	108 hmm_file = os.path.join(os.path.split(sys.argv[0])[0],
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	109 "whisson_et_al_rxlr_eer_cropped.hmm")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	110 if not os.path.isfile(hmm_file):
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	111 stop_err("Missing HMM file for Whisson et al. (2007)")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	112 if not get_hmmer_version(hmmer_search, "HMMER 2.3.2 (Oct 2003)"):
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	113 stop_err("Missing HMMER 2.3.2 (Oct 2003) binary, %s" % hmmer_searcher)
16 7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	114
6 a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	115 hmm_hits = set()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	116 valid_ids = set()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	117 for title, seq in fasta_iterator(fasta_file):
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	118 name = title.split(None,1)[0]
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	119 if name in valid_ids:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	120 stop_err("Duplicated identifier %r" % name)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	121 else:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	122 valid_ids.add(name)
16 7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	123 if not valid_ids:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	124 #Special case, don't need to run HMMER if there are no sequences
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	125 pass
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	126 else:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	127 #I've left the code to handle HMMER 3 in situ, in case
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	128 #we revisit the choice to insist on HMMER 2.
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	129 hmmer3 = (3 == get_hmmer_version(hmmer_search))
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	130 #Using zero (or 5.6?) for bitscore threshold
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	131 if hmmer3:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	132 #The HMMER3 table output is easy to parse
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	133 #In HMMER3 can't use both -T and -E
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	134 cmd = "%s -T 0 --tblout %s --noali %s %s > /dev/null" \
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	135 % (hmmer_search, hmm_output_file, hmm_file, fasta_file)
6 a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	136 else:
16 7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	137 #For HMMER2 we are stuck with parsing stdout
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	138 #Put 1e6 to effectively have no expectation threshold (otherwise
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	139 #HMMER defaults to 10 and the calculated e-value depends on the
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	140 #input FASTA file, and we can loose hits of interest).
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	141 cmd = "%s -T 0 -E 1e6 %s %s > %s" \
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	142 % (hmmer_search, hmm_file, fasta_file, hmm_output_file)
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	143 return_code = os.system(cmd)
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	144 if return_code:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	145 stop_err("Error %i from hmmsearch:\n%s" % (return_code, cmd), return_code)
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	146
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	147 handle = open(hmm_output_file)
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	148 for line in handle:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	149 if not line.strip():
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	150 #We expect blank lines in the HMMER2 stdout
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	151 continue
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	152 elif line.startswith("#"):
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	153 #Header
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	154 continue
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	155 else:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	156 name = line.split(None,1)[0]
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	157 #Should be a sequence name in the HMMER3 table output.
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	158 #Could be anything in the HMMER2 stdout.
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	159 if name in valid_ids:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	160 hmm_hits.add(name)
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	161 elif hmmer3:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	162 stop_err("Unexpected identifer %r in hmmsearch output" % name)
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	163 handle.close()
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	164 #if hmmer3:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	165 # print "HMMER3 hits for %i/%i" % (len(hmm_hits), len(valid_ids))
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	166 #else:
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	167 # print "HMMER2 hits for %i/%i" % (len(hmm_hits), len(valid_ids))
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	168 #print "%i/%i matched HMM" % (len(hmm_hits), len(valid_ids))
7de64c8b258d Uploaded v0.2.5, MIT licence, RST for README, citation information, development moved to GitHub peterjc parents: 6 diff changeset	169 os.remove(hmm_output_file)
6 a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	170 del valid_ids
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	171
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	172
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	173 #Prepare short list of candidates containing RXLR to pass to SignalP
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	174 assert min_rxlr_start > 0, "Min value one, since zero based counting"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	175 count = 0
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	176 total = 0
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	177 handle = open(signalp_input_file, "w")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	178 for title, seq in fasta_iterator(fasta_file):
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	179 total += 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	180 name = title.split(None,1)[0]
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	181 match = re_rxlr.search(seq[min_rxlr_start-1:].upper())
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	182 if match and min_rxlr_start - 1 + match.start() + 1 <= max_rxlr_start:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	183 #This is a potential RXLR, depending on the SignalP results.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	184 #Might as well truncate the sequence now, makes the temp file smaller
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	185 if signalp_trunc:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	186 handle.write(">%s (truncated)\n%s\n" % (name, seq[:signalp_trunc]))
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	187 else:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	188 #Does it matter we don't line wrap?
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	189 handle.write(">%s\n%s\n" % (name, seq))
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	190 count += 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	191 handle.close()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	192 #print "Running SignalP on %i/%i potentials." % (count, total)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	193
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	194
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	195 #Run SignalP (using our wrapper script to get multi-core support etc)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	196 signalp_script = os.path.join(os.path.split(sys.argv[0])[0], "signalp3.py")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	197 if not os.path.isfile(signalp_script):
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	198 stop_err("Error - missing signalp3.py script")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	199 cmd = "python %s euk %i %s %s %s" % (signalp_script, signalp_trunc, threads, signalp_input_file, signalp_output_file)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	200 return_code = os.system(cmd)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	201 if return_code:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	202 stop_err("Error %i from SignalP:\n%s" % (return_code, cmd))
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	203 #print "SignalP done"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	204
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	205 def parse_signalp(filename):
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	206 """Parse SignalP output, yield tuples of ID, HMM_Sprob_score and NN predicted signal peptide length.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	207
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	208 For signal peptide length we use NN_Ymax_pos (minus one).
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	209 """
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	210 handle = open(filename)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	211 line = handle.readline()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	212 assert line.startswith("#ID\t"), line
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	213 for line in handle:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	214 parts = line.rstrip("\t").split("\t")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	215 assert len(parts)==20, repr(line)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	216 yield parts[0], float(parts[18]), int(parts[5])-1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	217 handle.close()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	218
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	219
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	220 #Parse SignalP results and apply the strict RXLR criteria
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	221 total = 0
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	222 tally = dict()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	223 handle = open(tabular_file, "w")
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	224 handle.write("#ID\t%s\n" % model)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	225 signalp_results = parse_signalp(signalp_output_file)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	226 for title, seq in fasta_iterator(fasta_file):
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	227 total += 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	228 rxlr = "N"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	229 name = title.split(None,1)[0]
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	230 match = re_rxlr.search(seq[min_rxlr_start-1:].upper())
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	231 if match and min_rxlr_start - 1 + match.start() + 1 <= max_rxlr_start:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	232 del match
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	233 #This was the criteria for calling SignalP,
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	234 #so it will be in the SignalP results.
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	235 sp_id, sp_hmm_score, sp_nn_len = signalp_results.next()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	236 assert name == sp_id, "%s vs %s" % (name, sp_id)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	237 if sp_hmm_score >= min_signalp_hmm and min_sp <= sp_nn_len <= max_sp:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	238 match = re_rxlr.search(seq[sp_nn_len:].upper())
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	239 if match and match.start() + 1 <= max_sp_rxlr: #1-based counting
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	240 rxlr_start = sp_nn_len + match.start() + 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	241 if min_rxlr_start <= rxlr_start <= max_rxlr_start:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	242 rxlr = "Y"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	243 if model == "Whisson2007":
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	244 #Combine the signalp with regular expression heuristic and the HMM
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	245 if name in hmm_hits and rxlr == "N":
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	246 rxlr = "hmm" #HMM only
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	247 elif rxlr == "N":
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	248 rxlr = "neither" #Don't use N (no)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	249 elif name not in hmm_hits and rxlr == "Y":
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	250 rxlr = "re" #Heuristic only
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	251 #Now have a four way classifier: Y, hmm, re, neither
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	252 #and count is the number of Y results (both HMM and heuristic)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	253 handle.write("%s\t%s\n" % (name, rxlr))
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	254 try:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	255 tally[rxlr] += 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	256 except KeyError:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	257 tally[rxlr] = 1
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	258 handle.close()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	259 assert sum(tally.values()) == total
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	260
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	261 #Check the iterator is finished
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	262 try:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	263 signalp_results.next()
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	264 assert False, "Unexpected data in SignalP output"
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	265 except StopIteration:
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	266 pass
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	267
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	268 #Cleanup
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	269 os.remove(signalp_input_file)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	270 os.remove(signalp_output_file)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	271
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	272 #Short summary to stdout for Galaxy's info display
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	273 print "%s for %i sequences:" % (model, total)
a290c6d4e658 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository peterjc parents: diff changeset	274 print ", ".join("%s = %i" % kv for kv in sorted(tally.iteritems()))

Mercurial > repos > peterjc > tmhmm_and_signalp

annotate tools/protein_analysis/rxlr_motifs.py @ 16:7de64c8b258d draft