CCTOP

1python cctop_template.py -i tm.fa -s yes -f yes

  1#!/usr/bin/python
  2from __future__ import division
  3import os
  4import sys
  5import time
  6from lxml import etree
  7import lxml.etree as ET			
  8from os import popen
  9from os import system
 10sys.stdout = os.fdopen(sys.stdout.fileno(), 'w', 0) 
 11URL="http://cctop.enzim.ttk.mta.hu/php/"
 12def msg1(line):
 13	sys.stderr.write("'"+str(sys.argv[0])+"  -i fasta file' or '"+str(sys.argv[0])+"  -i fasta file [signal peptide] [tmfilter]'"+"\n")
 14	sys.stderr.write("Example: "+str(sys.argv[0])+" -i /home/user/protein.fas -s yes -f no"+"\n")
 15	sys.stderr.write("Mandatory option"+"\n")
 16	sys.stderr.write("-i filename or --input=filename: input sequences in fasta format"+"\n")
 17	sys.stderr.write("Other options:"+"\n")
 18	sys.stderr.write("-o filename or --output=filename: path to output results file, if omitted stdout is used"+"\n")
 19	sys.stderr.write("-s yes or --signal=yes: switch on signal peptid prediction (default is no signal peptid prediction)"+"\n")
 20	sys.stderr.write("-f yes or --tmfilter=yes: switch on tmfilter discrimination (default is no tmfilter discrimination)"+"\n")
 21	sys.stderr.write("After submission, the script will check for results every 30 sec."+"\n")
 22	sys.stderr.write("Status changes are displayed."+"\n")
 23def getargs(line):
 24	signalr="no"
 25	tmfilterr="no"
 26	loc=""
 27	out=""
 28	outfile=""
 29	for i in range (0, len(line)):
 30		if line[i][:2]=="-i":
 31			if i+1>=len(line):
 32				msg1(line)
 33				return ["",False,False,"none"]
 34			loc=line[i+1]		
 35		if line[i][:7]=="--input":
 36			if i+1>=len(line):
 37				msg1(line)
 38				return ["",False,False,"none"]
 39			loc=line[i][line[i].find("=")+1:]
 40		if line[i][:2]=="-s":
 41			if i+1>=len(line):
 42				msg1(line)
 43				return ["",False,False,"none"]
 44			signalr=line[i+1]
 45		if line[i][:8]=="--signal":
 46			if i+1>=len(line):
 47				msg1(line)
 48				return ["",False,False,"none"]
 49			signalr=line[i][line[i].find("=")+1:]
 50		if line[i][:2]=="-f":
 51			if i+1>=len(line):
 52				msg1(line)
 53				return ["",False,False,"none"]
 54			tmfilterr=line[i+1]
 55		if line[i][:10]=="--tmfilter":
 56			if i+1>=len(line):
 57				msg1(line)
 58				return ["",False,False,"none"]
 59			tmfilterr=line[i][line[i].find("=")+1:]
 60		if line[i][:2]=="-o":
 61			if i+1>=len(line):
 62				msg1(line)
 63				return ["",False,False,"none"]
 64			out=line[i+1]
 65		if line[i][:8]=="--output":
 66			if i+1>=len(line):
 67				msg1(line)
 68				return ["",False,False,"none"]
 69			out=line[i][line[i].find("=")+1:]
 70	if loc=="":
 71		msg1(line)
 72		return ["",False,False,"none"]
 73	else:
 74		try:
 75			fastafile=open(loc,"r")
 76			if out!="":
 77				try:
 78					outfile=open(out,"w")
 79				except IOError:
 80					sys.stderr.write("Invalid path, check output file location (use full path) or permissions"+"\n")
 81					sys.stderr.write(str(out)+"\n")				
 82					fastafile.close()
 83					return ["",False,False,"none"]
 84			if signalr.lower()=="yes":
 85				signal=True
 86				sys.stderr.write("Signal peptide prediction ON"+"\n")
 87			elif signalr.lower()=="no":
 88				signal=False
 89				sys.stderr.write("Signal peptide prediction OFF"+"\n")
 90			else:
 91				msg1(line)
 92				return ["",False,False,"none"]
 93			if tmfilterr.lower()=="yes":
 94				tmfilter=True
 95				sys.stderr.write("TMFilter ON"+"\n")
 96			elif tmfilterr.lower()=="no":
 97				tmfilter=False
 98				sys.stderr.write("TMFilter OFF"+"\n")
 99			else:
100				msg1(line)
101				return ["",False,False,"none"]
102		except IOError:
103			sys.stderr.write("Invalid path, check fasta file location (use full path)"+"\n")
104			sys.stderr.write(str(loc)+"\n")	
105			return ["",False,False,"none"]
106	return [fastafile,signal,tmfilter,outfile]
107def get(xml,jobID,doneID,error,running,scheduled):
108	for key in jobID:
109		check=True
110		for i in range (0, len(error)):
111			if error[i]==key:
112				check=False
113		for i in range (0, len(doneID)):
114			if doneID[i]==key:
115				check=False 
116		if check==True:
117			status=popen("wget -qO- '"+URL+"poll.php?jobId="+jobID[key]+"'")
118			content=status.readlines()
119			for i in range (0, len(content)):
120				if content[i].lower().find("invalid")!=-1:
121					error.append(key)
122				if content[i].lower().find("scheduled")!=-1:
123					sch=True
124					for j in range (0, len(scheduled)):
125						if scheduled[j]==key:
126							sch=False
127					if sch==True:
128						scheduled.append(key)
129				if content[i].lower().find("error")!=-1:
130					error.append(key)
131				if content[i].lower().find("finished")!=-1:
132					doneID.append(key)
133					result=popen("wget '"+URL+"result.php?jobId="+jobID[key]+"' -o /dev/null -O -")
134					content2=result.read()
135					doc = etree.fromstring(content2)
136					xml.append(doc)
137					try:
138						scheduled.remove(key)
139					except ValueError:
140						try:
141							running.remove(key)
142						except ValueError:
143							pass
144				if content[i].lower().find("running")!=-1:
145					r=True
146					for j in range (0, len(running)):
147						if running[j]==key:
148							r=False
149					if r==True:
150						running.append(key)
151					try:
152						scheduled.remove(key)
153					except ValueError:
154						pass
155			time.sleep(0.5)
156	return [xml,doneID,error,running,scheduled]
157def submit(sequences,signal,tmfilter):	
158	jobID={}
159	fail=[]
160	if tmfilter==True and signal==True:
161		for key in sequences:
162			up=popen("wget -qO- --post-data='sequence=%3E"+key+"%0A"+sequences[key]+"&tmFilter=&signalPred=' "+URL+"submit.php")
163			content=up.readlines()
164			if content[0].find("Invalid sequence")!=-1:
165				fail.append(key)
166			else:
167				ID=content[0].replace("ID: ","")
168				jobID[key]=ID.strip()
169			time.sleep(1)
170	elif signal==True:
171		for key in sequences:
172			up=popen("wget -qO- --post-data='sequence=%3E"+key+"%0A"+sequences[key]+"&signalPred=' "+URL+"submit.php")
173			content=up.readlines()
174			if content[0].find("Invalid sequence")!=-1:
175				fail.append(key)
176			else:
177				ID=content[0].replace("ID: ","")
178				jobID[key]=ID.strip()
179			time.sleep(1)
180	elif tmfilter==True:
181		for key in sequences:
182			up=popen("wget -qO- --post-data='sequence=%3E"+key+"%0A"+sequences[key]+"&tmFilter=' "+URL+"submit.php")
183			content=up.readlines()
184			if content[0].find("Invalid sequence")!=-1:
185				fail.append(key)
186			else:
187				ID=content[0].replace("ID: ","")
188				jobID[key]=ID.strip()
189			time.sleep(1)
190	else:	
191		for key in sequences:
192			up=popen("wget -qO- --post-data='sequence=%3E"+key+"%0A"+sequences[key]+"' "+URL+"submit.php")
193			content=up.readlines()
194			if content[0].find("Invalid sequence")!=-1:
195				fail.append(key)
196			else:
197				ID=content[0].replace("ID: ","")
198				jobID[key]=ID.strip()
199			time.sleep(1)
200	if len(fail)!=0:
201		sys.stderr.write("invalid sequence format, use 20 letter amino acid code"+"\n")
202		for i in range (0, len(fail)):
203			sys.stderr.write(fail[i]+"\n")
204	return [jobID,len(fail)]
205def read(fastafile):
206	seq=""
207	header=""
208	sequences={}
209	while 1:
210		line=fastafile.readline()
211		if line=="":
212			break
213		if line[0]==">":
214			if seq!="" and header!="":
215				seq=seq.replace(" ","")
216				sequences[header]=seq
217				seq=""
218			header=line[1:].strip()
219		else:
220			seq=seq+line.strip()
221	seq=seq.replace(" ","")
222	sequences[header]=seq
223	fastafile.close()
224	return sequences
225xml=[]
226if len(sys.argv)<2:
227	msg1(sys.argv)
228elif sys.argv[1].lower()=="help" or sys.argv[1].lower()=="-h" or sys.argv[1].lower()=="-help":
229	msg1(sys.argv)
230else:
231	[fastafile,signal,tmfilter,outfile]=getargs(sys.argv)
232	if fastafile!="" and outfile!="none":
233		sys.stderr.write("Reading fasta file"+"\n")
234		sequences=read(fastafile)
235		sys.stderr.write("Submitting sequences"+"\n")
236		[jobID,fail]=submit(sequences,signal,tmfilter)
237		doneID=[]
238		error=[]
239		running=[]
240		scheduled=[]
241		state=[-1,-1,-1,fail]
242		time.sleep(10)
243		if len(jobID)>0:
244			while 1:
245				sys.stderr.write("Checking results"+"\n")
246				[xml,doneID,error,running,scheduled]=get(xml,jobID,doneID,error,running,scheduled)
247				if len(doneID)==len(jobID):
248					sys.stderr.write("Status:"+"\n")
249					sys.stderr.write(str(len(doneID))+" job is done"+"\n")
250					sys.stderr.write(str(len(running))+" job is running"+"\n")
251					sys.stderr.write(str(len(scheduled))+" job is scheduled"+"\n")
252					if len(error)+fail!=0:
253						sys.stderr.write("Error in "+str(len(error)+fail)+" cases"+"\n")
254					sys.stderr.write("Done"+"\n")
255					break
256				elif len(sequences)==len(error)+fail+len(doneID):
257					sys.stderr.write("Status:"+"\n")
258					sys.stderr.write(str(len(doneID))+" job is done"+"\n")
259					sys.stderr.write(str(len(running))+" job is running"+"\n")
260					sys.stderr.write(str(len(scheduled))+" job is scheduled"+"\n")
261					if len(error)+fail!=0:
262						sys.stderr.write("Error in "+str(len(error)+fail)+" cases"+"\n")	
263					sys.stderr.write("Done"+"\n")
264					break
265				else:
266					if state!=[len(doneID),len(running),len(scheduled),len(error)+fail]:
267						sys.stderr.write("status:"+"\n")
268						sys.stderr.write(str(len(doneID))+" job is done"+"\n")
269						sys.stderr.write(str(len(running))+" job is running"+"\n")
270						sys.stderr.write(str(len(scheduled))+" job is scheduled"+"\n")
271						if len(error)+fail!=0:
272							sys.stderr.write("Error in "+str(len(error)+fail)+" cases"+"\n")				
273						sys.stderr.write("Please wait"+"\n")
274					else:
275						sys.stderr.write("No change in status"+"\n")
276					state=[len(doneID),len(running),len(scheduled),len(error)+fail]
277				time.sleep(30)
278		else:
279			sys.stderr.write("done"+"\n")
280	if outfile=="":
281		if len(xml)>0:
282			print "<CCTOPItems>"
283			for i in range (0, len(xml)):
284				doc=xml[i]
285				print(ET.tostring(doc, pretty_print=True).strip())
286			print "</CCTOPItems>"
287	elif outfile!="none":
288		outfile.write("<CCTOPItems>"+"\n")
289		for i in range (0, len(xml)):
290			doc=xml[i]
291			outfile.write(ET.tostring(doc, pretty_print=True).strip())
292			outfile.write("\n")
293		outfile.write("</CCTOPItems>"+"\n")
294		outfile.close()
295
296