#!/usr/bin/env python3
"""
optparse is deprecated in favor of argparse as of Python 2.7. However,
since 2.7 is not always present on many systems, at this writing,
it is safer to stick with optparse for now. It should be easy
to change later, since the syntax is very similar between argparse and optparse.
from optparse import OptionParser
"""
from optparse import OptionParser
import datetime
import getpass
import os
import re
import shutil
import stat
import subprocess
import sys
import time
'''
bl_soap-trans.py - Run SOAPdenovo-Trans
Synopsis: bl_soap-trans.py --tsv tsvfile [options]
@modified: April 6, 2019
@author: Brian Fristensky
@contact: Brian.Fristensky@umanitoba.ca
'''
#blib = os.environ.get("BIRCHPYLIB")
#sys.path.append(blib)
#from birchlib import Birchmod
PROGRAM = "bl_soap-trans.py : "
USAGE = "\n\tUSAGE: bl_soap-trans.py --tsv tsvfile [options]"
DEBUG = True
if DEBUG :
print('bl_soap-trans: Debugging mode on')
#BM = Birchmod(PROGRAM, USAGE)
# - - - - - - - - - - - - - Utility classes - - - - - - - - - - - - - - - - -
def chmod_ar(filename):
"""
Make a file world-readable.
"""
if os.path.exists(filename):
st = os.stat(filename)
os.chmod(filename, st.st_mode | stat.S_IREAD \
| stat.S_IRGRP | stat.S_IROTH)
def chmod_arx(filename):
"""
Make a file or directory world-readable and world-executable/searchable.
"""
if os.path.exists(filename):
st = os.stat(filename)
os.chmod(filename, st.st_mode | stat.S_IEXEC | stat.S_IREAD \
| stat.S_IXGRP | stat.S_IRGRP | stat.S_IXOTH \
| stat.S_IROTH)
def LocalHostname():
"""
Return the name of the local machine. Tries a number of methods
to get a name other than 'localhost' or a null result.
"""
import socket
import platform
def CheckName(name) :
if name == None or name.startswith("localhost") or name == "" :
OKAY = False
else :
OKAY = True
return OKAY
name = os.getenv('HOSTNAME')
if not CheckName(name) :
name = platform.uname()[1]
if not CheckName(name) :
if socket.gethostname().find('.')>=0:
name=socket.gethostname()
else:
name=socket.gethostbyaddr(socket.gethostname())[0]
return name
def SendEmail(From,To,Subject,Text) :
"""
Very simple email method adapted from:
http://stackoverflow.com/questions/882712/sending-html-email-using-python
There are more elaborate examples on this site for sending
HTML messages and attachments.
"""
import smtplib
from email.mime.multipart import MIMEMultipart
from email.mime.text import MIMEText
Host = 'localhost'
msg = MIMEMultipart('alternative')
msg['Subject'] = Subject
Html = """\
%s
""" %(Text)
part1 = MIMEText(Text, 'plain')
part2 = MIMEText(Html, 'html')
msg.attach(part1)
msg.attach(part2)
try:
server = smtplib.SMTP(Host)
server.sendmail(From, To, msg.as_string())
server.quit()
print("Successfully sent email")
except :
print("Error: unable to send email")
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
class Parameters:
"""
Wrapper class for command line parameters
"""
def __init__(self):
"""
Initializes arguments:
TSVFILE = ""
PROGRAM = "SOAPdenovo-Trans-31mer"
MERGELEVEL = 1
MINCONTIG = 100
MAXREADLEN = 100
ASMFLAGS = 3
PREFIX = ""
OUTDIR = ""
EMAIL = ""
Then calls read_args() to fill in their values from command line
"""
self.TSVFILE = "."
self.PROGRAM = "SOAPdenovo-Trans-31mer"
self.THREADS = "2"
self.MERGELEVEL = 1
self.MINCONTIG = 100
self.MAXREADLEN = 100
self.ASMFLAGS = 3
self.OUTDIR = "."
self.PREFIX = ""
self.EMAIL = ""
self.read_args()
if DEBUG :
print('------------ Parameters from command line ------')
print(' TSVFILE: ' + self.TSVFILE)
print(' PROGRAM: ' + self.PROGRAM)
print(' MERGELEVEL: ' + str(self.MERGELEVEL))
print(' MINCONTIG: ' + str(self.MINCONTIG))
print(' MAXREADLEN: ' + str(self.MAXREADLEN))
print(' ASMFLAGS: ' + str(self.ASMFLAGS))
print(' THREADS: ' + str(self.THREADS))
print(' OUTDIR: ' + self.OUTDIR)
print(' PREFIX: ' + self.PREFIX)
print(' EMAIL: ' + self.EMAIL)
print()
def read_args(self):
"""
Read command line arguments into a Parameter object
"""
parser = OptionParser()
parser.add_option("--tsvfile", dest="tsvfile", action="store", default="",
help="list of paired/unpaired read files")
parser.add_option("--program", dest="program", action="store", default="SOAPdenovo-Trans-31mer",
help="SOAPdenovo-Trans-31mer or SOAPdenovo-Trans-127mer")
parser.add_option("--mergelevel", dest="mergelevel", action="store", default=1,
help="strength of merging similar sequences")
parser.add_option("--mincontig", dest="mincontig", action="store", default=100,
help="shortest contig for scaffolding")
parser.add_option("--maxreadlen", dest="maxreadlen", action="store", default=100,
help="max. read length")
parser.add_option("--asmflags", dest="asmflags", action="store", default=3,
help="when to use reads: contigs|scaffolds|both")
parser.add_option("--threads", dest="threads", action="store", default=2,
help="number of threads to use")
#parser.add_option("--buckets", dest="buckets", action="store_true", default=False,
# help="Write number of sequences per bucket to output")
parser.add_option("--outdir", dest="outdir", action="store", default=".",
help="output directory")
parser.add_option("--prefix", dest="prefix", action="store", default="",
help="prefix for output files")
parser.add_option("--email", dest="email", action="store", default="",
help="send email upon completion")
(options, args) = parser.parse_args()
self.TSVFILE = options.tsvfile
self.PROGRAM = options.program
self.MERGELEVEL = options.mergelevel
self.MINCONTIG = options.mincontig
self.MAXREADLEN = options.maxreadlen
self.ASMFLAGS = options.asmflags
self.THREADS = options.threads
self.OUTDIR = options.outdir
self.PREFIX = options.prefix
self.EMAIL = options.email
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
class TSVFiles :
"""
Methods for reading lists of paired read TSV files, and for
writing lists to output.
"""
def __init__(self):
"""
Initializes arguments:
READPAIRS = []
"""
self.READPAIRS = []
def ReadTSVfile(self,FN) :
"""
TSV file containing names of paired-end and/or single end read files.
Paired-end files are on lines such as
leftreadfile.fq.gzrightreadfile.fq.gz
Single-end files have a each file on a separate line
reads1.fq.gz
reads2.fq.gz
reads3.fq.gz
"""
TAB = '\t'
F = open(FN,"r")
for line in F.readlines() :
line = line.strip()
if len(line) > 0 and not line.startswith('#') :
# get rid of double quotes that enclose fields when some programs write
# output, and then split by TABs.
tokens = line.replace('"','').split(TAB)
# ignore blank fields. Add either single or pair of filenames
# to list. Only process names from first two fields on a line
# and ignore other fields.
if len(tokens) > 0 :
r1 = tokens[0].strip()
if len(r1) > 0 :
fnames = [r1]
else :
fnames = []
if len(tokens) > 1 :
r2 = tokens[1].strip()
if len(r2) > 0 :
fnames.append(r2)
if len(fnames) > 0 :
self.READPAIRS.append(fnames)
if DEBUG :
print(str(self.READPAIRS))
F.close()
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
# Create config file for running SOAPdenovo-Trans
def BuildConfig(P,TF,CFN) :
# Return a code indicating the type of file(s), based on file exension(s)
def FileType(PR) :
FT = ""
Fastq = ['.fq','.fastq']
# There are too many file extensions that have been used for fasta files,
# so we assume that it it's not a fastq file, it must be a fasta file.
if len(PR) == 2 :
Ext1 = os.path.splitext(PR[0])[1].lower()
Ext2 = os.path.splitext(PR[1])[1].lower()
if Ext1 in Fastq and Ext2 in Fastq :
FT = 'qn'
else :
FT = 'fn'
else :
Ext1 = os.path.splitext(PR[0])[1].lower()
if Ext1 in Fastq :
FT = 'q'
else :
FT = 'f'
return FT
NL = '\n'
CFILE = open(CFN,'w')
CFILE.write('#===== SOAPdenovo-Trans Config file) generated by bl_soap-trans.py =====' + NL)
CFILE.write(NL)
CFILE.write('# - - - - - PARAMETERS - - - - -' + NL)
CFILE.write('max_rd_len=' + P.MAXREADLEN + NL)
# Multiple libraries are not supported, but we still need this line
# or SOAP crashes.
CFILE.write('[LIB]' + NL)
CFILE.write('asm_flags=' + str(P.ASMFLAGS) + NL)
CFILE.write(NL)
CFILE.write('# - - - - - READ FILES - - - - -' + NL)
for PR in TF :
FT = FileType(PR)
if FT == 'qn' : # paired fastq files
CFILE.write('q1=' + PR[0] + NL)
CFILE.write('q2=' + PR[1] + NL)
elif FT == 'fn' : # paired fasta files
CFILE.write('f1=' + PR[0] + NL)
CFILE.write('f2=' + PR[1] + NL)
elif FT == 'q' : # single fastq file
CFILE.write('q=' + PR[0] + NL)
elif FT == 'f' : # single fasta file
CFILE.write('f=' + PR[0] + NL)
CFILE.close()
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
def RunSOAPdenovoTrans(P,PR,CFN,LOGFILE) :
COMMAND=[P.PROGRAM]
COMMAND.extend(["all", "-s", CFN, "-p", P.THREADS, "-M", P.MERGELEVEL, "-L", P.MINCONTIG, "-o", P.PREFIX])
# Run the command - - - - - - - - - - - - - - - - -
LOGFILE.write('======== SOAPdenovo-Trans ==========' + '\n')
LOGFILE.write(str(COMMAND) + '\n')
StartTime = datetime.datetime.now()
LOGFILE.write('Start time: ' + str(StartTime) + '\n')
LOGFILE.write('\n')
LOGFILE.flush()
p = subprocess.Popen(COMMAND,stdout=LOGFILE, stderr=LOGFILE)
p.wait()
FinishTime = datetime.datetime.now()
LOGFILE.write('\n')
LOGFILE.write('Finish time: ' + str(FinishTime) + '\n')
ElapsedTime = FinishTime - StartTime
LOGFILE.write('Elapsed time: ' + str(ElapsedTime) + '\n')
#======================== MAIN PROCEDURE ==========================
def main():
"""
Called when not in documentation mode.
"""
# Read parameters from command line
P = Parameters()
TF = TSVFiles()
if not P.TSVFILE == "" :
TF.ReadTSVfile(P.TSVFILE)
OKAY = True
# We'll be working in the output directory, so we need
# absolute file paths to the input files.
for PR in TF.READPAIRS :
PR[0] = os.path.abspath(PR[0])
if len(PR) == 2 :
PR[1] = os.path.abspath(PR[1])
else :
OKAY = False
# From now on, everything happens in the output directory.
if not os.path.exists(P.OUTDIR) :
os.mkdir(P.OUTDIR)
os.chdir(P.OUTDIR)
LOGFN = 'bl_soap-trans.log'
# First create the file, then re-open it so that we can append to it.
# We first create with 'w' just to make sure that we're creating a fresh copy,
# rather than appending to an existing copy from a previous run.
LOGFILE = open(LOGFN,'w')
LOGFILE.close()
LOGFILE = open(LOGFN,'a')
LOGFILE.write('\n')
if not OKAY :
LOGFILE.write('bl_soap-trans.py: >>> Need to specify a list of sequencing read files <<<' + '\n')
# Create config file for running SOAPdenovo-Trans
CFN = P.PREFIX + '.config'
BuildConfig(P,TF.READPAIRS,CFN)
# Run SOAPdenovo-Trans
if OKAY :
RunSOAPdenovoTrans(P,PR,CFN,LOGFILE)
LOGFILE.close()
# Notify user when job is done, if email address was
# supplied using --email
if P.EMAIL != "" :
Sender = getpass.getuser() + '@' + LocalHostname()
Subject = 'bl_soap-trans.py completed'
Message = 'bl_soap-trans.py: Completed
'
LOGFILE = open(os.path.join('bl_soap-trans.log'),'r')
for line in LOGFILE.readlines() :
Message = Message + line + '
'
LOGFILE.close()
SendEmail(Sender,[P.EMAIL],Subject,Message)
if __name__ == "__main__":
main()
#else:
#used to generate documentation
# import doctest
# doctest.testmod()
#if (BM.documentor() or "-test" in sys.argv):
# pass
#else:
# main()