osiris_phylogenetics: phylogenies/phytab_raxml

author	osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
date	Tue, 11 Mar 2014 12:19:13 -0700
parents
children

rev	line source
0 5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	1 #!/usr/bin/env python
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	2
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	3 ## This tool runs RAxML's parsimony inference on a phytab input.
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	4 ## If N = # of nodes requested in job runner, then N RAxML jobs will run simultaneously. Make sure that the
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	5 ## number of processors ('ppn') in the job runner matches the 'numthreads' commandline argument -T.
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	6 ##
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	7 ## Usage: ./phytab_raxml_using_ptree.parallel.py -i <phytabinput> -e <model> -f <modelfile> -T 4
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	8 ## example: ./phytab_raxml_using_ptree.parallel.py -i myphytab.txt -e PROTGAMMAWAG -f None -T 4
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	9 ## or: ./phytab_raxml_using_ptree.parallel.py -i myphtab.txt -e None -f modelsforeachpartition.txt -T 4
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	10 ##
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	11 ## outputs a tab-delimited file with gene-partition and newick parsimony tree on each line.
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	12
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	13 import optparse
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	14 import os
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	15 import subprocess
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	16 import multiprocessing
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	17
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	18 RESULTS_DIR = 'results'
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	19 RESULTS_FILE = 'parsimony_results.txt'
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	20 RAXML_PREFIX = 'RAxML_parsimonyTree.'
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	21
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	22 def unescape(string):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	23 mapped_chars = {
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	24 '>': '__gt__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	25 '<': '__lt__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	26 "'": '__sq__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	27 '"': '__dq__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	28 '[': '__ob__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	29 ']': '__cb__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	30 '{': '__oc__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	31 '}': '__cc__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	32 '@': '__at__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	33 '\n': '__cn__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	34 '\r': '__cr__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	35 '\t': '__tc__',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	36 '#': '__pd__'
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	37 }
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	38
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	39 for key, value in mapped_chars.iteritems():
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	40 string = string.replace(value, key)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	41
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	42 return string
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	43
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	44
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	45 class Species:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	46 def __init__(self, string):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	47 lis = string.split('\t')
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	48 # print lis
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	49 self.species = lis[0]
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	50 self.gene = lis[1]
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	51 self.name = lis[2]
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	52 self.sequence = lis[3]
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	53
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	54 def toString(self):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	55 return self.species + '\t' + self.sequence
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	56
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	57
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	58 class Gene:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	59 def __init__(self, name):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	60 self.name = name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	61 self.count = 0
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	62 self.length = 0
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	63 self.species = []
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	64
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	65 def output(self):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	66 file_name = self.name + ".phy"
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	67 location = RESULTS_DIR + os.sep + file_name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	68 with open(location, 'w') as f:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	69 f.write(str(self.count) + '\t' + str(self.length) + '\n')
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	70 for s in self.species:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	71 f.write(s.toString())
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	72 return file_name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	73
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	74 def add(self, species):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	75 if species.name == "":
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	76 return
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	77 self.species.append(species)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	78 self.count += 1
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	79 if self.length == 0:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	80 self.length = len(species.sequence) - 1
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	81
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	82
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	83 def output_species(species):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	84 file_name = species.gene + ".phy"
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	85 location = RESULTS_DIR + os.sep + file_name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	86 with open(location, 'a') as f:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	87 f.write(species.toString())
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	88 return file_name
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	89
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	90
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	91 def process_phytab(input):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	92 files = set()
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	93 genes = dict()
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	94 with open(input) as f:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	95 for line in f:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	96 if len(line) < 4:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	97 continue
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	98 species = Species(line)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	99 if species.gene in genes:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	100 genes[species.gene].add(species)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	101 else:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	102 gene = Gene(species.gene)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	103 gene.add(species)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	104 genes[gene.name] = gene
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	105 for k, gene in genes.iteritems():
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	106 files.add(gene.output())
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	107 return files
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	108
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	109
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	110 def runRaxml(list_of_files, evo, evoDict,NUMTHREADS):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	111 for gene_file in list_of_files:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	112 if gene_file.split(".")[0] in evoDict:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	113 newEvo = evoDict[gene_file.split(".")[0]]
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	114 else:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	115 newEvo = evo
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	116 # cpu_count = str(multiprocessing.cpu_count())
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	117 file_name = RESULTS_DIR + os.sep + gene_file
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	118 # to run parsimony trees:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	119 popen = subprocess.Popen(['raxmlHPC-PTHREADS', '-T', cpu_count,'-f', 'd', '-s', file_name,'-y', '-m', newEvo, '-n', gene_file[:-4]+'.tre', '-p', '34'])
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	120 # to run likelihood trees:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	121 # popen = subprocess.Popen(['raxmlHPC-PTHREADS', "-T", NUMTHREADS, "-s", file_name, '-m', newEvo, '-n', gene_file[:-4], '-p', '34'])
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	122 popen.wait()
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	123
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	124
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	125 def toData(text, name):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	126 text = name + "\t" + text.replace("\n", "\\n")
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	127 return text
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	128
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	129 def readEfile(efile):
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	130 evoDict = {}
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	131 with open(efile, "r") as f:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	132 for line in f:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	133 pair = line.split("\t")
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	134 evoDict[pair[0].strip()] = pair[1].strip()
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	135 return evoDict
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	136
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	137 def main():
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	138 usage = """%prog [options]
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	139 options (listed below) default to 'None' if omitted
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	140 """
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	141 parser = optparse.OptionParser(usage=usage)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	142
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	143 parser.add_option(
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	144 '-i', '--in',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	145 dest='input',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	146 action='store',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	147 type='string',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	148 metavar="FILE",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	149 help='Name of input data.')
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	150
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	151 parser.add_option(
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	152 '-e', '--evo',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	153 dest='evo',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	154 action='store',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	155 type='string',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	156 metavar="EVO",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	157 help='Evolution model.')
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	158
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	159 parser.add_option(
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	160 '-f', '--evo-file',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	161 dest='efile',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	162 action='store',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	163 type='string',
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	164 metavar="EVO_FILE",
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	165 help='Evolution model file. Format is gene_name [tab] evolution_model.')
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	166
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	167 parser.add_option('-T', '--numthread',dest='numthreads', action='store',type='int', metavar="NUMT", help='Provide number of threads for RAxML')
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	168 options, args = parser.parse_args()
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	169
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	170 os.mkdir(RESULTS_DIR)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	171
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	172 list_of_species_files = process_phytab(unescape(options.input))
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	173
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	174
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	175 try:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	176 evoDict = readEfile(unescape(options.efile))
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	177 except IOError:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	178 print "Could not find evolution model file, using:", unescape(options.evo)
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	179 evoDict = {}
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	180
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	181 runRaxml(list_of_species_files, unescape(options.evo), evoDict,str(options.numthreads))
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	182
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	183 result = [file for file in os.listdir('./') if file.startswith(RAXML_PREFIX)]
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	184 with open(RESULTS_DIR + os.sep + RESULTS_FILE, "w") as f:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	185 for file in result:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	186 with open(file, "r") as r:
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	187 f.write(file[len(RAXML_PREFIX):-4] + '\t' + r.read())
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	188
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	189 if __name__ == '__main__':
5b9a38ec4a39 First commit of old repositories osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> parents: diff changeset	190 main()

0

5b9a38ec4a39 First commit of old repositories