Corrected a bug of overwriting on the fasta file

computePred.R

@@ -42,6 +42,13 @@ write.table(res[[3]][[3]],paste0(prot,"_pssm80.txt"))
 # read evolutionary traces computed by JET
 jet=read.table(paste(prot,"_jet.res",sep=""),head=TRUE)
 if(sum(colnames(jet)=="traceMax")==1){trace=jet[,"traceMax"]}else{trace=jet[,"trace"]}
+
+#You should comment line 44 to use this functionality. Or maybe, it should go into normalize functions
+#That was what we originally decided with Alessandra. 
+#To get max values of PC, CV or Trace
+#trace = c()
+#for (row in 1:nrow(jet)) { trace<-append(trace, max(jet[row, "trace"], jet[row, "pc"])) }
+
 #traceAli = sweep(binAli, MARGIN=2, trace, `*`)
 # compute evolutionary distances of all sequences with respect to the query
 distTrace = binAli[2:N[1],] %*% trace^2

default.conf

@@ -20,6 +20,7 @@ url		http://www.rcsb.org/pdb/downloadFile.do		URL of PDB server
 >Filter
 min_identity	0.20						min sequence identity
 max_identity	0.98						max sequence identity
+
 *****************************************
 >Sample
 length_cutoff	0.8						minimum sequence length expressed in number of residues	
@@ -28,12 +29,13 @@ length_cutoff	0.8						minimum sequence length expressed in number of residues
 >Software
 clustalW	/usr/local/bin/clustalw2			clustalW system dependent command
 muscle          /usr/bin/muscle                                 muscle system dependent command
-naccess		/opt/JET2/naccess2.1.1/naccess			naccess system dependent command
-psiblast	/opt/blast-2.2.27+/bin/psiblast 		psiblast system dependent command     
+naccess		/home/tekpinar/research/carbone-lab-software/naccess2.1.1/naccess 	naccess system dependent command
+psiblast	/usr/bin/psiblast				psiblast system dependent command
+
 *****************************************
 >Data
-substMatrix	/opt/JET2/matrix				directory location of matrices used in JET (Blosum62, gonnet and hsdm)
-blastDatabases	/disk1/blastdb/					directory location of databases used for local blast (nr{0-7})
+substMatrix	/home/tekpinar/research/carbone-lab-software/JET2/matrix	directory location of matrices used in JET (Blosum62, gonnet and hsdm)
+blastDatabases	/opt/blastdb					directory location of databases used for local blast (nr{0-7})
 
 *****************************************
 >ET
@@ -56,9 +58,9 @@ max_dist        20.0                                            max distance
 >Interface
 cutoff		0         					minimum percentage accessible surface variation of an interface residu
 ligand		no						(yes|no) keep contact of ligand (SUBSTRATE, PRODUCT and COFACTOR of database ENZYME) to compute interface of protein
-enzymeCpd	/opt/JET/jet/data/enzyme.txt			location of file containing database ENZYME
+enzymeCpd	/home/tekpinar/research/carbone-lab-software/JET2/jet/data/enzyme.txt			location of file containing database ENZYME
 homologousPDB	no						(yes|no) add interface residues of homologous structures (find in pdb database clustered at 95% of identities) to interface of protein
-clusteredPDB	/opt/JET/jet/data/clusters95.txt		location of pdb database clustered at 95% of identities
+clusteredPDB	/home/tekpinar/research/carbone-lab-software/JET2/jet/data/clusters95.txt		location of pdb database clustered at 95% of identities
 
 *****************************************
 >Cluster

gemme.py

@@ -132,9 +132,15 @@ def doit(inAli,mutFile,retMet,bFile,fFile,n,N):
     doit(args.input,args.mutations,args.retrievingMethod,args.blastFile,args.fastaFile)
     """
     prot,seq,nl=extractQuerySeq(inAli)
+
     createPDB(prot,seq)
+ 
     print "query protein: "+prot
     print "computing conservation levels..."
+
+    #I intend to run JET2 completely externally!!
+    #It is too much buggy and it has too many dependencies. 
+    #Using it with a Docker or Singularity may be the best solution!
     launchJET(prot,retMet,bFile,fFile,n,N,nl)
     print "done"
     launchPred(prot,inAli,mutFile)

gemmeAnal.py

@@ -10,6 +10,7 @@ import argparse
 import re
 import math
 import subprocess
+import shutil
 
 
 # Extract the query sequence from the input alignment
@@ -67,21 +68,64 @@ def editConfJET(N):
 
 # Run JET to compute TJET values
 def launchJET(prot,retMet,bFile,fFile,n,N,nl):
+	"""
+		Call JET2 and produce prot+"_jet.res" file.
+
+		prot+"_jet.res" will be used in the following steps (in launchPred)
+		to calculate independent and epistatic models.
+		Ideally, this call to JET2 should be from Dockers or Singularity 
+		because installing all requirements of JET2 is a pain in the ass!
+
+	Parameters
+    ----------
+    prot: string ???
+        Name of the protein ???
+    retMet: string
+		Retreival method of multiple sequence alignments file
+		It can be 'input', 'local' or 'server'. Default is local.
+	bFile: string 
+        A multiple sequence alignment file obtained with psiblast.
+		It is used only if the retMet (explained above) is input.
+	fFile: string 
+        A multiple sequence alignment file obtained with psiblast.
+		It is used only if the retMet (explained above) is input.
+	n: int
+		Number of JET2 iterations.
+	N: int
+		Default 20000
+	nl: int
+		Number of lines after > character in the query sequences file.
+		It is obtained in extractQuerySeq() function.
+
+    Returns
+    -------
+	Nothing 
+	"""
 
 	subprocess.call("cp $GEMME_PATH/default.conf .",shell=True)
 	if retMet=="input":
 		if bFile!='':
 			subprocess.call("cp "+bFile+" "+prot+"_A.psiblast",shell=True)
-			jetcmd = "java -Xmx1000m -cp $JET_PATH:$JET_PATH/jet/extLibs/vecmath.jar jet.JET -c default.conf -i "+prot+".pdb -o `pwd` -p J -r input -b "+prot+"_A.psiblast -d chain -n "+n+" > "+prot+".out"
+			jetcmd = "java -Xmx1000m -cp $JET2_PATH:$JET2_PATH/jet/extLibs/vecmath.jar jet.JET -c default.conf -i "+prot+".pdb -o `pwd` -p J -r input -b "+prot+"_A.psiblast -d chain -n "+n+" > "+prot+".out"
 		else:
                         print N
 			editConfJET(N)
+
+			if(fFile == prot+"_A.fasta"):
+				shutil.copy2(fFile, fFile+".orig")
+				#I think this subprocess call causes overwriting of the fasta file. 
+				#subprocess.call("cp "+fFile+" "+prot+"_A.fasta",shell=True)
+				grpcmd="grep -m "+str(int(N)+1)+" -A "+str(nl)+" '^>' "+fFile+".orig > "+prot+"_A.fasta"
+			else:
 				#subprocess.call("cp "+fFile+" "+prot+"_A.fasta",shell=True)
-			subprocess.call("grep -m "+str(int(N)+1)+" -A "+str(nl)+" '^>' "+fFile+" > "+prot+"_A.fasta",shell=True)
-			jetcmd = "java -Xmx1000m -cp $JET_PATH:$JET_PATH/jet/extLibs/vecmath.jar jet.JET -c default.conf -i "+prot+".pdb -o `pwd` -p J -r input -f "+prot+"_A.fasta -d chain -n "+n+" > "+prot+".out"
+				grpcmd="grep -m "+str(int(N)+1)+" -A "+str(nl)+" '^>' "+fFile+" > "+prot+"_A.fasta"
+			
+			print("\nRunning:\n"+grpcmd)
+			subprocess.call(grpcmd,shell=True)
+			jetcmd = "java -Xmx1000m -cp $JET2_PATH:$JET2_PATH/jet/extLibs/vecmath.jar jet.JET -c default.conf -i "+prot+".pdb -o `pwd` -p J -r input -f "+prot+"_A.fasta -d chain -n "+n+" > "+prot+".out"
 	else:
-		jetcmd = "java -Xmx1000m -cp $JET_PATH:$JET_PATH/jet/extLibs/vecmath.jar jet.JET -c default.conf -i "+prot+".pdb -o `pwd` -p J -r "+retMet+" -d chain -n "+n+" > "+prot+".out"
-	print jetcmd
+		jetcmd = "java -Xmx1000m -cp $JET2_PATH:$JET2_PATH/jet/extLibs/vecmath.jar jet.JET -c default.conf -i "+prot+".pdb -o `pwd` -p J -r "+retMet+" -d chain -n "+n+" > "+prot+".out"
+	print("\nRunning:\n"+jetcmd)
 	reCode=subprocess.call(jetcmd,shell=True)
 	if os.path.isfile(prot+"/"+prot+"_jet.res"):
 		os.rename(prot+"/"+prot+"_jet.res",prot+"_jet.res")
@@ -94,8 +138,11 @@ def launchPred(prot,inAli,mutFile):
 		rcmd="Rscript --save $GEMME_PATH/computePred.R "+prot+" "+inAli+" FALSE "+mutFile
 	else:
 		rcmd="Rscript --save $GEMME_PATH/computePred.R "+prot+" "+inAli+" TRUE none"
+
 	print rcmd
 	reCode=subprocess.call(rcmd,shell=True)
+
+	#Add plots here with gemmemore
 	return(reCode)
 
 # Remove temporary files
@@ -113,7 +160,7 @@ def cleanTheMess(prot,bFile,fFile):
 				os.remove(prot+"_A.fasta")		
 #	if os.path.isfile(prot+"/"+prot+"_jet.res"):
 #		os.rename(prot+"/"+prot+"_jet.res",prot+"_jet.res")
-	os.remove(prot+".pdb")
+#	os.remove(prot+".pdb")
 	dir_name = prot+"/"
 	if os.path.isdir(dir_name):
 		for f in os.listdir(dir_name):