vhm.py in big_simon-0.1.0

- old
+ new

@@ -44,12 +44,13 @@
 
 ## log file ##
 filelog = open(os.path.join(tmpDir, 'vhm.log'), 'w')
 
 ## name length ##
-nameLen = 93 - len(options.outDir)
+# nameLen = 93 - len(options.outDir)
 #### possibly because of the kmercount folder name for each contig is too long?
+nameLen = 99999999999999
 
 
 #################### 0: preparation ############################
 
 ## path to the programs
@@ -209,11 +210,11 @@
 	sys.stdout.write("WARNING: no hostTaxa file provided, creating a dummy one \n")
 	filelog.write("WARNING: no hostTaxa file provided, creating a dummy one \n")
 	hostTaxaFileWrite = open(hostTaxaFile, 'w') ## make file blank
 	hostTaxaFileWrite.close()
 	hostTaxaFileWrite = open(hostTaxaFile, 'a')
-	
+
 	hostTaxaFileWrite.write("hostNCBIName	hostName	hostSuperkingdom	hostPhylum	hostClass	hostOrder	hostFamily	hostGenus	hostSpecies\n")
 	for currentFileName in hostFaList :
 		if currentFileName.startswith('.') :
 			continue
 		if len(currentFileName) > nameLen :
@@ -233,11 +234,11 @@
 else :
 	hostTaxaFile = os.path.join(options.outDir, os.path.basename(options.hostTaxaFile)+"_new.txt")
 	hostTaxaTable = numpy.genfromtxt(options.hostTaxaFile,delimiter="\t", dtype=str)
 	hostTaxaTable[hostTaxaTable=='']='unknown'
 	numpy.savetxt(hostTaxaFile, hostTaxaTable, fmt="%s", delimiter='\t', newline='\n')
-    
+
 filelog.flush()
 
 #################### 1: count kmer and prepare list files ############################
 #sys.stdout.write("Step 1: counting kmers \n")
 
@@ -257,11 +258,11 @@
 		currentFileNameS = currentFileName
 	sys.stdout.write("Step 1: counting kmers for virus " + currentFileNameS + "\n")
 	filelog.write("Step 1: counting kmers for virus " + currentFileNameS + "\n")
 	for w in range(1, (kmax+1)) :
 		currentFilePath = os.path.join(options.virusFaDir, currentFileName)
-    
+
 		currentKmerCountPath = os.path.join(kmerCountPath, currentFileNameS)
 		cmdKmer = countKmerOut + " -l -k " + str(w) + \
 										" -i " + currentFilePath +\
 										" -o " + currentKmerCountPath +\
 										" -s " + currentFileNameS
@@ -276,13 +277,13 @@
 													str(2) + "\n")
 	else :
 		sys.stderr.write( "ERROR in counting kmers for " + currentFileNameS + "\n")
 		filelog.write( "ERROR in counting kmers for " + currentFileNameS + "\n")
 		sys.exit(0)
-  
+
 	filelog.flush()
-  
+
 	end_time = time.time()
 	count += 1
 	#sys.stdout.write(str(end_time - start_time) + "s for " + str(count) + " seqs \n")
 	sys.stdout.write(" (Average time for counting kmers for one virus: " + str(round((end_time - start_time)/count,4)) + "s) \n" )
 	sys.stdout.write(" (ETR for counting kmers for viruses: " + str(round((end_time - start_time)/count*(len(virusFaList)-count), 4)) + "s) \n")
@@ -327,13 +328,13 @@
 												str(2) + "\n")
 	else :
 		sys.stderr.write( "ERROR in counting kmers for " + currentFileNameS + "\n")
 		filelog.write( "ERROR in counting kmers for " + currentFileNameS + "\n")
 		sys.exit(0)
-  
+
 	filelog.flush()
-  
+
 	end_time = time.time()
 	count += 1
 	#sys.stdout.write(str(end_time - start_time) + "s for " + str(count) + " seqs \n")
 	sys.stdout.write(" (Average time for counting kmers for one host: " + str(round((end_time - start_time)/count, 4)) + "s) \n")
 	sys.stdout.write(" (ETR for counting kmers for hosts: " + str(round((end_time - start_time)/count*(len(hostFaList)-count), 4)) + "s) \n")
@@ -364,10 +365,10 @@
 #for c in iter(lambda: cmdCptMeasureOut.stderr.read(1), ''):
 for c in iter(cmdCptMeasureOut.stderr.readline, b''):
 	sys.stdout.write(c.decode("utf-8"))
 	filelog.write(c.decode("utf-8"))
 	filelog.flush()
-  
+
 	end_time = time.time()
 	count += 1
 	sys.stdout.write("   (Average time for computing dissimilarities for one virus-host pair: " + str(round((end_time - start_time)/count/len(virusFaList), 4)) + "s) \n")
 	sys.stdout.write("   (ETR for computing dissimilarities for virus-host pairs: " + str(round((end_time - start_time)/count*(len(hostFaList)-count), 4)) + "s) \n")
 	filelog.write("   (Average time for computing dissimilarities for one virus-host pair: " + str(round((end_time - start_time)/count/len(virusFaList), 4)) + "s) \n")