Fixed bug causing CodeML to hang

Author: icwells

bin/04_CallCodeML.py

@@ -17,6 +17,7 @@
 import math
 import os
 import re
+from callPhyML import phyml
 
 # Define max number of threads and devnull for capturing stdout
 MAXCPU = cpu_count()
@@ -49,39 +50,34 @@ def controlFiles(indir, outdir, forward, cpu):
 		os.mkdir(tmp)
 	except FileExistsError:
 		pass
-	cmd = ("python bin/04_callPhyML.py -i" + indir + " -o " + tmp + 
-			" -t " + str(cpu))
-	if forward:
-		cmd += " -f " + forward
-	phyml = Popen(split(cmd))
-	phyml.wait()
-	#if phyml.returncode() == 0:
-	return True
+	multiple = phyml(indir, tmp, cpu, forward)
+	return True, multiple
 
 #-----------------------------------------------------------------------------
 
-def runCodeml(ap, outdir, finished, completed, gene):
+def runCodeml(ap, outdir, finished, completed, multiple, gene):
 	'''Creates temporary control and tree files and runs CodeML.'''
 	filename = gene.split("/")[-1]
 	geneid = filename.split(".")[0]
 	wd = outdir + "tmp/" + geneid + "/"
-	if filename.split(".")[1] == "2":
-		if len(glob(wd + "*")) > 1:
-			# Skip pairwise genes if tree files are present
-			pass
-	elif (geneid + "\n") in completed:
+	if (geneid + "\n") in completed:
 		pass
 	else:
-		tempctl = wd + geneid + ".ctl"
-		# Calls CodeML
+		tempctl = wd + "codeml.ctl"
 		os.chdir(wd)
-		cm = Popen(split(ap + "/paml/bin/codeml " + tempctl), 
-						stdout = DEVNULL)
-		cm.wait()
-		if cm.returncode == 0:
-			# Append gene ID to list of finishedCodeML.txt
-			with open(finished, "a") as fin:
-				fin.write(geneid + "\n")
+		if multiple == True:
+			if filename.split(".")[1] == "2":
+				pass
+			else:
+				# Calls CodeML if 3 or more sequences are present
+				cm = Popen(split(ap + "paml/bin/codeml " + tempctl), 
+								stdout = DEVNULL)
+		elif multiple == False:
+			# Call CodeML for all files
+			cm = Popen(split(ap + "paml/bin/codeml " + tempctl), 
+							stdout = DEVNULL)
+		with open(finished, "a") as fin:
+			fin.write(geneid + "\n")
 
 #-----------------------------------------------------------------------------
 
@@ -115,7 +111,7 @@ def main():
 	cleanup = args.cleanUp
 	# Reads in required data
 	finished, completed = outputFiles(outdir)
-	run = controlFiles(indir, outdir, forward, cpu)
+	run, multiple = controlFiles(indir, outdir, forward, cpu)
 	if run == True:
 		# Call CodeML after PhyML completes.
 		genes = glob(indir + "*.phylip")
@@ -126,15 +122,15 @@ def main():
 		elif l > cpu:
 			chunk = int(math.ceil(l/cpu))
 		pool = Pool(processes = cpu)
-		func = partial(runCodeml, ap, outdir, finished, completed)
-		print("\tRunning CodeML with", str(cpu), "threads....")
-		rcml = pool.imap(func, genes, chunksize = chunk)
+		func = partial(runCodeml, ap, outdir, finished, completed, multiple)
+		print(("\tRunning CodeML with {0!s} threads....").format(cpu))
+		rcml = pool.imap_unordered(func, genes, chunksize = chunk)
 		pool.close()
-		pool.join()	
+		pool.join()		
 	# Remove tmp directory
 	if cleanup == True:
 		shutil.rmtree(outdir + "tmp/")
-	print("\tCodeML runtime: ", datetime.now() - starttime, "\n")
+	print(("\tCodeML runtime: {0!s}").format(datetime.now() - starttime))
 
 if __name__ == "__main__":
 	main()

bin/04_callPhyML.py bin/callPhyML.pybin/04_callPhyML.py renamed to bin/callPhyML.py

@@ -67,7 +67,7 @@ def makeTree(indir, outdir, forward, completed, ctl, gene):
 			pass
 		# Set unique file names
 		outfile = (out + geneid + "." + filename.split(".")[1] + ".mlc")
-		tempctl = wd + geneid + ".ctl"
+		tempctl = wd + "codeml.ctl"
 		treefile = wd + filename + "_phyml_tree.txt"
 		# Make unique control file
 		makeCtl(gene, outfile, tempctl, treefile, ctl)
@@ -130,7 +130,7 @@ def pairwiseControl(indir, outdir, ctl, gene):
 		out += i + "/"
 	# Set unique file names
 	outfile = (out + geneid +"."+ filename.split(".")[1] + ".mlc")
-	tempctl = wd + geneid + ".ctl"
+	tempctl = wd + "codeml.ctl"
 	treefile = ""
 	# Make unique control file
 	makeCtl(gene, outfile, tempctl, treefile, ctl)
@@ -150,27 +150,9 @@ def makeCtl(gene, outfile, tempctl, treefile, ctl):
 
 #-----------------------------------------------------------------------------
 
-def main():
-	# Parse command
-	parser = argparse.ArgumentParser(description="Runs CodeML on all files \
-in a directory.")
-	parser.add_argument("-i", help="Path to input directory.")
-	parser.add_argument("-o", help="Path to temp directory.")
-	parser.add_argument("-t", type=int, default=1, help="Number of threads.")
-	parser.add_argument("-f", default="", 
-help="Forward species (name must be the same as it appears in input files.")
-	args = parser.parse_args()
-	# Assign arguments
-	indir = args.i
-	if indir[-1] != "/":
-		indir += "/"
-	outdir = args.o
-	if outdir[-1] != "/":
-		outdir += "/"
-	cpu = args.t
-	forward = args.f
+def phyml(indir, outdir, cpu, forward=""):
 	# Reads in required data
-	completed, ctl, multiple = outputFiles( outdir)
+	completed, ctl, multiple = outputFiles(outdir)
 	# Call PhyML for multiple alignments or write pairwise control files.
 	genes = glob(indir + "*.phylip")
 	l = int(len(genes))
@@ -196,6 +178,4 @@ def main():
 		rpml = pool.imap(func, genes, chunksize = chunk)
 		pool.close()
 		pool.join()	
-
-if __name__ == "__main__":
-	main()
+	return multiple