amplicon_analysis_pipeline: amplicon_analysis

annotate amplicon_analysis_pipeline.py @ 17:e1ba19551395 draft

planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty

author	pjbriggs
date	Tue, 03 Jul 2018 07:20:12 -0400
parents	4bd582d23d4e
children	877d53e2d963

rev	line source
0 b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	1 #!/usr/bin/env python
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	2 #
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	3 # Wrapper script to run Amplicon_analysis_pipeline.sh
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	4 # from Galaxy tool
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	5
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	6 import sys
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	7 import os
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	8 import argparse
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	9 import subprocess
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	10 import glob
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	11
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	12 class PipelineCmd(object):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	13 def __init__(self,cmd):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	14 self.cmd = [str(cmd)]
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	15 def add_args(self,*args):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	16 for arg in args:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	17 self.cmd.append(str(arg))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	18 def __repr__(self):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	19 return ' '.join([str(arg) for arg in self.cmd])
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	20
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	21 def ahref(target,name=None,type=None):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	22 if name is None:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	23 name = os.path.basename(target)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	24 ahref = "<a href='%s'" % target
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	25 if type is not None:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	26 ahref += " type='%s'" % type
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	27 ahref += ">%s</a>" % name
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	28 return ahref
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	29
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	30 def check_errors():
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	31 # Errors in Amplicon_analysis_pipeline.log
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	32 with open('Amplicon_analysis_pipeline.log','r') as pipeline_log:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	33 log = pipeline_log.read()
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	34 if "Names in the first column of Metatable.txt and in the second column of Final_name.txt do not match" in log:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	35 print_error("""* Sample IDs don't match dataset names *
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	36
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	37 The sample IDs (first column of the Metatable file) don't match the
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	38 supplied sample names for the input Fastq pairs.
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	39 """)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	40 # Errors in pipeline output
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	41 with open('pipeline.log','r') as pipeline_log:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	42 log = pipeline_log.read()
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	43 if "Errors and/or warnings detected in mapping file" in log:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	44 with open("Metatable_log/Metatable.log","r") as metatable_log:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	45 # Echo the Metatable log file to the tool log
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	46 print_error("""* Error in Metatable mapping file *
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	47
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	48 %s""" % metatable_log.read())
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	49 elif "No header line was found in mapping file" in log:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	50 # Report error to the tool log
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	51 print_error("""* No header in Metatable mapping file *
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	52
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	53 Check you've specified the correct file as the input Metatable""")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	54
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	55 def print_error(message):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	56 width = max([len(line) for line in message.split('\n')]) + 4
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	57 sys.stderr.write("\n%s\n" % (''width))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	58 for line in message.split('\n'):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	59 sys.stderr.write("* %s%s \n" % (line,' '(width-len(line)-4)))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	60 sys.stderr.write("%s\n\n" % (''width))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	61
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	62 def clean_up_name(sample):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	63 # Remove trailing "_L[0-9]+_001" from Fastq
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	64 # pair names
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	65 split_name = sample.split('_')
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	66 if split_name[-1] == "001":
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	67 split_name = split_name[:-1]
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	68 if split_name[-1].startswith('L'):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	69 try:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	70 int(split_name[-1][1:])
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	71 split_name = split_name[:-1]
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	72 except ValueError:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	73 pass
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	74 return '_'.join(split_name)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	75
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	76 def list_outputs(filen=None):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	77 # List the output directory contents
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	78 # If filen is specified then will be the filename to
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	79 # write to, otherwise write to stdout
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	80 if filen is not None:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	81 fp = open(filen,'w')
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	82 else:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	83 fp = sys.stdout
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	84 results_dir = os.path.abspath("RESULTS")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	85 fp.write("Listing contents of output dir %s:\n" % results_dir)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	86 ix = 0
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	87 for d,dirs,files in os.walk(results_dir):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	88 ix += 1
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	89 fp.write("-- %d: %s\n" % (ix,
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	90 os.path.relpath(d,results_dir)))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	91 for f in files:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	92 ix += 1
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	93 fp.write("---- %d: %s\n" % (ix,
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	94 os.path.relpath(f,results_dir)))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	95 # Close output file
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	96 if filen is not None:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	97 fp.close()
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	98
8 8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	99 def find_executable(name):
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	100 # Locate executable on PATH
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	101 for p in os.environ['PATH'].split(os.pathsep):
12 fb2af52d67d1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit 53544cd620f81c5cfbdcb277e835f1c6da8588b0-dirty pjbriggs parents: 11 diff changeset	102 exe = os.path.join(p,name)
9 32a43bc3dd58 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 8 diff changeset	103 if os.path.isfile(exe) and os.access(exe,os.X_OK):
8 8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	104 return exe
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	105 return None
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	106
0 b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	107 if __name__ == "__main__":
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	108 # Command line
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	109 print "Amplicon analysis: starting"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	110 p = argparse.ArgumentParser()
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	111 p.add_argument("metatable",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	112 metavar="METATABLE_FILE",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	113 help="Metatable.txt file")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	114 p.add_argument("fastq_pairs",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	115 metavar="SAMPLE_NAME FQ_R1 FQ_R2",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	116 nargs="+",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	117 default=list(),
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	118 help="Triplets of SAMPLE_NAME followed by "
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	119 "a R1/R2 FASTQ file pair")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	120 p.add_argument("-g",dest="forward_pcr_primer")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	121 p.add_argument("-G",dest="reverse_pcr_primer")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	122 p.add_argument("-q",dest="trimming_threshold")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	123 p.add_argument("-O",dest="minimum_overlap")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	124 p.add_argument("-L",dest="minimum_length")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	125 p.add_argument("-l",dest="sliding_window_length")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	126 p.add_argument("-P",dest="pipeline",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	127 choices=["vsearch","uparse","qiime"],
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	128 type=str.lower,
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	129 default="vsearch")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	130 p.add_argument("-S",dest="use_silva",action="store_true")
2 a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	131 p.add_argument("-H",dest="use_homd",action="store_true")
0 b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	132 p.add_argument("-r",dest="reference_data_path")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	133 p.add_argument("-c",dest="categories_file")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	134 args = p.parse_args()
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	135
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	136 # Build the environment for running the pipeline
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	137 print "Amplicon analysis: building the environment"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	138 metatable_file = os.path.abspath(args.metatable)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	139 os.symlink(metatable_file,"Metatable.txt")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	140 print "-- made symlink to Metatable.txt"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	141
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	142 # Link to Categories.txt file (if provided)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	143 if args.categories_file is not None:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	144 categories_file = os.path.abspath(args.categories_file)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	145 os.symlink(categories_file,"Categories.txt")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	146 print "-- made symlink to Categories.txt"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	147
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	148 # Link to FASTQs and construct Final_name.txt file
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	149 sample_names = []
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	150 with open("Final_name.txt",'w') as final_name:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	151 fastqs = iter(args.fastq_pairs)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	152 for sample_name,fqr1,fqr2 in zip(fastqs,fastqs,fastqs):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	153 sample_name = clean_up_name(sample_name)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	154 r1 = "%s_R1_.fastq" % sample_name
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	155 r2 = "%s_R2_.fastq" % sample_name
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	156 os.symlink(fqr1,r1)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	157 os.symlink(fqr2,r2)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	158 final_name.write("%s\n" % '\t'.join((r1,sample_name)))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	159 final_name.write("%s\n" % '\t'.join((r2,sample_name)))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	160 sample_names.append(sample_name)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	161
2 a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	162 # Reference database
a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	163 if args.use_silva:
a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	164 ref_database = "silva"
a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	165 elif args.use_homd:
a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	166 ref_database = "homd"
a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	167 else:
a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	168 ref_database = "gg"
a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	169
8 8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	170 # Executables
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	171 os.mkdir("bin")
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	172 os.environ["PATH"] = os.path.abspath("bin") + \
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	173 os.pathsep + \
10 b47cf1d6fb47 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 9 diff changeset	174 os.environ["PATH"]
8 8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	175 print "-- updated PATH: %s" % os.environ["PATH"]
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	176 # Pipeline wants 'vsearch113' but bioconda version is just
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	177 # 'vsearch'
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	178 vsearch = find_executable("vsearch113")
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	179 if vsearch is None:
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	180 vsearch = find_executable("vsearch")
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	181 if vsearch:
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	182 os.symlink(vsearch,os.path.join("bin","vsearch113"))
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	183 print "-- made symlink to %s" % vsearch
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	184 else:
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	185 sys.stderr.write("Missing 'vsearch[113]'\n")
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	186 # Pipeline wants 'fasta-splitter.pl' but bioconda provides
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	187 # 'fasta-splitter'
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	188 fasta_splitter = find_executable("fasta-splitter.pl")
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	189 if fasta_splitter is None:
11 570228bb2fb7 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit 53544cd620f81c5cfbdcb277e835f1c6da8588b0 pjbriggs parents: 10 diff changeset	190 fasta_splitter = os.readlink(
8 8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	191 find_executable("fasta-splitter"))
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	192 if fasta_splitter:
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	193 os.symlink(vsearch,os.path.join("bin","fasta-splitter.pl"))
13 c87b166cbfe1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit 777167fc30406b792d0cf924753069526f8b3e5e-dirty pjbriggs parents: 12 diff changeset	194 print "-- made symlink to %s" % fasta_splitter
8 8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	195 else:
8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	196 sys.stderr.write("Missing 'fasta-splitter[.pl]'\n")
15 07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	197 # Qiime 1.8.0 wants 'RDP_JAR_PATH' env var to point to
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	198 # the rdp_classifier.jar file
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	199 # biconda provides this in the 'share/rdp_classifier/' dir
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	200 rdp_classifier = find_executable("rdp_classifier")
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	201 if rdp_classifier:
16 4bd582d23d4e planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 15 diff changeset	202 rdp_classifier = os.path.normpath(
4bd582d23d4e planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 15 diff changeset	203 os.path.join(os.path.dirname(rdp_classifier),
15 07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	204 "..",
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	205 "share",
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	206 "rdp_classifier",
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	207 "rdp_classifier.jar"))
17 e1ba19551395 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 16 diff changeset	208 if rdp_classifier and os.path.exists(rdp_classifier):
16 4bd582d23d4e planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 15 diff changeset	209 os.environ["RDP_JAR_PATH"] = rdp_classifier
15 07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	210 print "-- set RDP_JAR_PATH: %s" % os.environ["RDP_JAR_PATH"]
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	211 else:
07b04a63abf1 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit af3afa7bd2c2441074879938cd93d9965908bbc7-dirty pjbriggs parents: 13 diff changeset	212 sys.stderr.write("Missing 'rdp_classifier.jar'\n")
8 8292f8262a9d planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit f72ff88ee11d09703bf38524dbb8026ee14bc3ef-dirty pjbriggs parents: 2 diff changeset	213
0 b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	214 # Construct the pipeline command
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	215 print "Amplicon analysis: constructing pipeline command"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	216 pipeline = PipelineCmd("Amplicon_analysis_pipeline.sh")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	217 if args.forward_pcr_primer:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	218 pipeline.add_args("-g",args.forward_pcr_primer)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	219 if args.reverse_pcr_primer:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	220 pipeline.add_args("-G",args.reverse_pcr_primer)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	221 if args.trimming_threshold:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	222 pipeline.add_args("-q",args.trimming_threshold)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	223 if args.minimum_overlap:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	224 pipeline.add_args("-O",args.minimum_overlap)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	225 if args.minimum_length:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	226 pipeline.add_args("-L",args.minimum_length)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	227 if args.sliding_window_length:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	228 pipeline.add_args("-l",args.sliding_window_length)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	229 if args.reference_data_path:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	230 pipeline.add_args("-r",args.reference_data_path)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	231 pipeline.add_args("-P",args.pipeline)
2 a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	232 if ref_database == "silva":
0 b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	233 pipeline.add_args("-S")
2 a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	234 elif ref_database == "homd":
a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	235 pipeline.add_args("-H")
0 b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	236
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	237 # Echo the pipeline command to stdout
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	238 print "Running %s" % pipeline
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	239
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	240 # Run the pipeline
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	241 with open("pipeline.log","w") as pipeline_out:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	242 try:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	243 subprocess.check_call(pipeline.cmd,
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	244 stdout=pipeline_out,
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	245 stderr=subprocess.STDOUT)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	246 exit_code = 0
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	247 print "Pipeline completed ok"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	248 except subprocess.CalledProcessError as ex:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	249 # Non-zero exit status
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	250 sys.stderr.write("Pipeline failed: exit code %s\n" %
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	251 ex.returncode)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	252 exit_code = ex.returncode
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	253 except Exception as ex:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	254 # Some other problem
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	255 sys.stderr.write("Unexpected error: %s\n" % str(ex))
2 a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	256 exit_code = 1
0 b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	257
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	258 # Write out the list of outputs
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	259 outputs_file = "Pipeline_outputs.txt"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	260 list_outputs(outputs_file)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	261
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	262 # Check for log file
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	263 log_file = "Amplicon_analysis_pipeline.log"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	264 if os.path.exists(log_file):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	265 print "Found log file: %s" % log_file
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	266 if exit_code == 0:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	267 # Create an HTML file to link to log files etc
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	268 # NB the paths to the files should be correct once
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	269 # copied by Galaxy on job completion
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	270 with open("pipeline_outputs.html","w") as html_out:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	271 html_out.write("""<html>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	272 <head>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	273 <title>Amplicon analysis pipeline: log files</title>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	274 <head>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	275 <body>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	276 <h1>Amplicon analysis pipeline: log files</h1>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	277 <ul>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	278 """)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	279 html_out.write(
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	280 "<li>%s</li>\n" %
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	281 ahref("Amplicon_analysis_pipeline.log",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	282 type="text/plain"))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	283 html_out.write(
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	284 "<li>%s</li>\n" %
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	285 ahref("pipeline.log",type="text/plain"))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	286 html_out.write(
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	287 "<li>%s</li>\n" %
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	288 ahref("Pipeline_outputs.txt",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	289 type="text/plain"))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	290 html_out.write(
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	291 "<li>%s</li>\n" %
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	292 ahref("Metatable.html"))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	293 html_out.write("""<ul>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	294 </body>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	295 </html>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	296 """)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	297 else:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	298 # Check for known error messages
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	299 check_errors()
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	300 # Write pipeline stdout to tool stderr
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	301 sys.stderr.write("\nOutput from pipeline:\n")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	302 with open("pipeline.log",'r') as log:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	303 sys.stderr.write("%s" % log.read())
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	304 # Write log file contents to tool log
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	305 print "\nAmplicon_analysis_pipeline.log:"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	306 with open(log_file,'r') as log:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	307 print "%s" % log.read()
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	308 else:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	309 sys.stderr.write("ERROR missing log file \"%s\"\n" %
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	310 log_file)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	311
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	312 # Handle FastQC boxplots
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	313 print "Amplicon analysis: collating per base quality boxplots"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	314 with open("fastqc_quality_boxplots.html","w") as quality_boxplots:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	315 # PHRED value for trimming
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	316 phred_score = 20
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	317 if args.trimming_threshold is not None:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	318 phred_score = args.trimming_threshold
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	319 # Write header for HTML output file
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	320 quality_boxplots.write("""<html>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	321 <head>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	322 <title>Amplicon analysis pipeline: Per-base Quality Boxplots (FastQC)</title>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	323 <head>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	324 <body>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	325 <h1>Amplicon analysis pipeline: Per-base Quality Boxplots (FastQC)</h1>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	326 """)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	327 # Look for raw and trimmed FastQC output for each sample
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	328 for sample_name in sample_names:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	329 fastqc_dir = os.path.join(sample_name,"FastQC")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	330 quality_boxplots.write("<h2>%s</h2>" % sample_name)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	331 for d in ("Raw","cutdapt_sickle/Q%s" % phred_score):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	332 quality_boxplots.write("<h3>%s</h3>" % d)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	333 fastqc_html_files = glob.glob(
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	334 os.path.join(fastqc_dir,d,"*_fastqc.html"))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	335 if not fastqc_html_files:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	336 quality_boxplots.write("<p>No FastQC outputs found</p>")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	337 continue
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	338 # Pull out the per-base quality boxplots
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	339 for f in fastqc_html_files:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	340 boxplot = None
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	341 with open(f) as fp:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	342 for line in fp.read().split(">"):
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	343 try:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	344 line.index("alt=\"Per base quality graph\"")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	345 boxplot = line + ">"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	346 break
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	347 except ValueError:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	348 pass
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	349 if boxplot is None:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	350 boxplot = "Missing plot"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	351 quality_boxplots.write("<h4>%s</h4><p>%s</p>" %
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	352 (os.path.basename(f),
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	353 boxplot))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	354 quality_boxplots.write("""</body>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	355 </html>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	356 """)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	357
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	358 # Handle additional output when categories file was supplied
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	359 if args.categories_file is not None:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	360 # Alpha diversity boxplots
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	361 print "Amplicon analysis: indexing alpha diversity boxplots"
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	362 boxplots_dir = os.path.abspath(
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	363 os.path.join("RESULTS",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	364 "%s_%s" % (args.pipeline.title(),
2 a898ee628343 Updated to version 1.2.1.0 pjbriggs parents: 0 diff changeset	365 ref_database),
0 b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	366 "Alpha_diversity",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	367 "Alpha_diversity_boxplot",
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	368 "Categories_shannon"))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	369 print "Amplicon analysis: gathering PDFs from %s" % boxplots_dir
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	370 boxplot_pdfs = [os.path.basename(pdf)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	371 for pdf in
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	372 sorted(glob.glob(
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	373 os.path.join(boxplots_dir,"*.pdf")))]
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	374 with open("alpha_diversity_boxplots.html","w") as boxplots_out:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	375 boxplots_out.write("""<html>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	376 <head>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	377 <title>Amplicon analysis pipeline: Alpha Diversity Boxplots (Shannon)</title>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	378 <head>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	379 <body>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	380 <h1>Amplicon analysis pipeline: Alpha Diversity Boxplots (Shannon)</h1>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	381 """)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	382 boxplots_out.write("<ul>\n")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	383 for pdf in boxplot_pdfs:
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	384 boxplots_out.write("<li>%s</li>\n" % ahref(pdf))
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	385 boxplots_out.write("<ul>\n")
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	386 boxplots_out.write("""</body>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	387 </html>
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	388 """)
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	389
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	390 # Finish
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	391 print "Amplicon analysis: finishing, exit code: %s" % exit_code
b433086738d6 planemo upload for repository https://github.com/pjbriggs/Amplicon_analysis-galaxy commit ba3e5b591407db52a586361efb21927c8171ec0e pjbriggs parents: diff changeset	392 sys.exit(exit_code)

Mercurial > repos > pjbriggs > amplicon_analysis_pipeline

annotate amplicon_analysis_pipeline.py @ 17:e1ba19551395 draft