-
Notifications
You must be signed in to change notification settings - Fork 19
/
ancestralReconstruction.py
executable file
·251 lines (229 loc) · 9.21 KB
/
ancestralReconstruction.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
#!/usr/bin/python
import sys, os, argparse, subprocess, shlex, glob
from datetime import datetime
from multiprocessing import cpu_count
from multiprocessing.dummy import Pool as ThreadPool
from Bio import SeqIO
from Bio.Seq import Seq
from Bio.SeqRecord import SeqRecord
from Bio.Alphabet import IUPAC
################################################################################
# This script finds core genes from a a list of genomes and a groups file in
# the format output by OrthoMCL. It aligns the core genes and makes a tree from
# the alignment. It uses Lazarus to run PAML ancestral reconstruction, and it
# concatenates the ancestral gene sequences.
#
# Program Requirements: translatorX, mafft, lazarus, raxml, biopython
# Input: OrthoMCL groups file, list of genomes (outgroup last), nucleotide
# sequences for genes in the genomes
################################################################################
TRANSLATOR_X_PATH = "/opt/PepPrograms/translatorx_vLocal.pl"
LAZARUS_PATH = "/opt/PepPrograms/project-lazarus/lazarus.py"
RAXML_PATH = "/opt/PepPrograms/standard-RAxML/raxmlHPC-PTHREADS-AVX"
class FullPaths(argparse.Action):
"""Expand user- and relative-paths"""
def __call__(self, parser, namespace, values, option_string=None):
setattr(namespace, self.dest,
os.path.abspath(os.path.expanduser(values)))
def listdir_fullpath(d):
return [os.path.join(d, f) for f in os.listdir(d)]
def is_dir(dirname):
"""Checks if a path is a directory"""
if not os.path.isdir(dirname):
msg = "{0} is not a directory".format(dirname)
raise argparse.ArgumentTypeError(msg)
else:
return dirname
def is_file(filename):
"""Checks if a file exists"""
if not os.path.isfile(filename):
msg = "{0} is not a file".format(filename)
raise argparse.ArgumentTypeError(msg)
else:
return filename
def get_args():
"""Parse command line arguments"""
parser = argparse.ArgumentParser(description='Ancestral reconstruction of\
core genome')
parser.add_argument("groups", help="OrthoMCL groups file", action=FullPaths,
type=is_file)
parser.add_argument("genomes", help="File listing genomes to be included in\
the analysis- outgroup last", action=FullPaths, type=is_file)
parser.add_argument("genes",
help="Directory with .fasta files of nucleotide sequences for genomes",
action=FullPaths, type=is_dir)
parser.add_argument("-t", "--threads",
help="Number of threads to use (default: 2)",
type=int, default=2, choices=range(2, cpu_count()))
return parser.parse_args()
def check_paths():
for i in [TRANSLATOR_X_PATH, LAZARUS_PATH, RAXML_PATH]:
if not os.path.isfile(i):
msg = "{0} does not exist".format(i)
print msg
sys.exit()
def call_with_log(cmd):
"""Calls a system command with the subprocess module. Redirects both stdout
and stderr to a log file"""
cmd = cmd.format(**(kvmap))
logfile = open(wd + current_datetime+".log", "a+")
logfile.write("Executing command: " + cmd + "\n")
logfile.flush()
ret = subprocess.call(shlex.split(cmd), stdout=logfile, stderr=logfile)
if(ret != 0):
print("Pipeline did not complete successfully. \n Command : \n\n" +
cmd + "\n\n returned with non-zero code: " + str(ret))
logfile.close()
def read_groups_file(inFileName):
""" Read in groups file and create dictionary of group name and proteins in
group"""
print "Reading groups file"
inFile = open(inFileName, 'r')
groupsDict = {}
for line in inFile:
line = line.strip()
entries = line.split(':')
groupName = entries[0]
groupProteins = entries[1][1:].split(' ')
groupsDict[groupName] = groupProteins
inFile.close()
print len(groupsDict)
return groupsDict
def get_core_genes(groupsDict, genomes):
""" Gets core genes for genomes in list """
coreGenes = set()
for group in groupsDict:
genomeList = []
proteinList = groupsDict[group]
for protein in proteinList:
ids = protein.split('|')
genomeID = ids[0]
genomeList.append(genomeID)
genomeSet = set(genomeList)
if set(genomes.keys()).issubset(genomeSet):
if len(genomeList) == len(genomeSet):
coreGenes.add(group)
print len(coreGenes)
return coreGenes
def make_unaligned_fasta(dnaDirectory, groupsDict, coreGenes, genomes, og):
""" Reads through files in provided directory to find gene sequences that
match the proteins in the groups dictionary"""
print "Collecting core genes"
def make_fasta(group):
proteins = groupsDict[group]
out = open(group + '/' + group + '.fasta', 'w')
records = []
outgroup_gene = og
ingroup_genes = []
for protein in proteins:
seqID = protein.split('|')[0]
if seqID in genomes:
protein = protein.split('|')[1]
newRec = seqRecordDict[protein]
newRec.description = ""
records.append(newRec)
if og in newRec.id:
outgroup_gene = newRec.id
else:
ingroup_genes.append(newRec.id)
SeqIO.write(records, out, 'fasta')
return (group, ingroup_genes, outgroup_gene)
files = listdir_fullpath(dnaDirectory)
seqRecordDict = {}
seqIDs = []
for f in files:
handle = open(f, 'r')
for record in SeqIO.parse(handle, 'fasta'):
seqRecordDict[record.id] = record
pool = ThreadPool(args.threads)
seqIDs = pool.map(make_fasta, coreGenes)
pool.close()
pool.join()
return seqIDs
def align_gene_sequences(coreGenes):
""" Use MAFFT to align gene sequences"""
print "Aligning core genes"
def run_translatorX(infile):
outfile = "%s/alignment/%s" % (os.path.dirname(infile),
os.path.splitext(os.path.split(infile)[1])[0])
call_with_log(TRANSLATOR_X_PATH + " -i %s -o %s -p F"
% (infile, outfile))
return outfile + ".nt_ali.fasta"
files = [x + "/" + x + ".fasta" for x in coreGenes]
pool = ThreadPool(args.threads)
outfileList = pool.map(run_translatorX, files)
pool.close()
pool.join()
def make_trees(coreGenes):
""" Use RAxML to calculate maximum likelihood phylogeny """
print "Running RAxML"
def run_raxml(coreGene):
alignFile = "%s/alignment/%s.nt_ali.fasta" % (coreGene, coreGene)
outdir = os.path.abspath("%s/tree/" % coreGene)
name = "ml_" + os.path.splitext(os.path.basename(alignFile))[0]
call_with_log(RAXML_PATH + " -T 2 -m GTRGAMMA -# 20 -p 123 -s %s -w %s \
-n %s" % (alignFile, outdir, name))
return name
pool = ThreadPool(args.threads/2)
outNames = pool.map(run_raxml, coreGenes)
pool.close()
pool.join()
def ancestral_reconstruction(outgroup_genes):
""" Use Lazarus wrapper to run paml ancestral reconstruction """
print "Ancestral Reconstruction"
def run_lazarus(outgroup_gene):
coreGene, ingroup, outgroup = outgroup_gene
align = "%s/alignment/%s.nt_ali.fasta" % (coreGene, coreGene)
tree = "%s/tree/RAxML_bestTree.ml_%s.nt_ali" % (coreGene, coreGene)
model = "/opt/PepPrograms/paml4.8/dat/wag.dat"
outdir = os.path.abspath("%s/ancestral/" % coreGene)
call_with_log(LAZARUS_PATH + " --codeml --outputdir %s --verbose 9 \
--alignment %s --tree %s --model %s --asrv 4 --gapcorrect --getanc --ingroup %s\
--outgroup %s" % (outdir, align, tree, model, "[%s]" % (",".join(ingroup)),
"[%s]" % (outgroup)))
pool = ThreadPool(args.threads)
pool.map(run_lazarus, outgroup_genes)
pool.close()
pool.join()
def concatenate_genes(coreGenes):
""" Using Biopython, concatenate ancestral genes into one file"""
print "Concatenating ancestral reconstructions"
def parse_lazarus_output(coreGene):
ancRecFile = open("%s/ancestral/ancestor.out.txt" % coreGene, "r")
for i,line in enumerate(ancRecFile):
if i == 13:
record = SeqRecord(Seq(line.strip(), IUPAC.ambiguous_dna),
id=coreGene, description="")
return record
pool = ThreadPool(args.threads)
recs = pool.map(parse_lazarus_output, coreGenes)
SeqIO.write(recs, "ancestralGenes.fa", "fasta")
check_paths()
args = get_args()
current_datetime = datetime.today().strftime("%d-%m-%Y-%H%M")
wd = os.getcwd() + "/"
kvmap = {'projectname':'coreAlignment'}
genomes = {}
orderedGenomes = []
with open(args.genomes, "r") as inFile:
for line in inFile:
genomes[line.strip()[-4:]] = line.strip()
orderedGenomes.append(line.strip())
og = orderedGenomes[-1]
groupsDict = read_groups_file(args.groups)
coreGenes = get_core_genes(groupsDict, genomes)
for n in coreGenes:
try:
os.mkdir(n)
os.mkdir(n + "/alignment")
os.mkdir(n + "/tree")
os.mkdir(n + "/ancestral")
except OSError as e:
if e.errno != os.errno.EEXIST:
raise
outgroup_genes = make_unaligned_fasta(args.genes, groupsDict, coreGenes,
genomes, og)
align_gene_sequences(coreGenes)
make_trees(coreGenes)
ancestral_reconstruction(outgroup_genes)
concatenate_genes(coreGenes)