Add files via upload

Makefile

@@ -0,0 +1,11 @@
+#make bash file executable
+#compile c++ code
+CXX = g++
+CXXFLAGS= -Wformat -O3
+
+TRate:
+	chmod +x TRate.sh
+
+	$(CXX) -o $@ $(CXXFLAGS) TR.v1.cpp
+
+clean:

README.md

@@ -0,0 +1,57 @@
+# _TRate_
+The _TRate_ program computes "rate" of each transcript according to given coverage file. Transcripts are coded by coordinates of their exons (bed file for now, gtf/gff in future). Rate is computed as total _mass_ of exons within the transcript divided by total _length_ of exons. _Mass_ is taken as approximation of the area under coverage curve, i.e. sum of areas of coverage rectangles, and length is computed as sum of lengths of bedgraph intervals within exons.
+
+The _TRate_ program takes in two arguments in fixed order.
+
+1. Exons_file - coordinate sorted bed file that provides locations of exons for the corresponding transcript provided in column 4.
+
+Exons_file format example
+
+     C0000570	10420	10640	Transcript1
+     C0000570	128078	128167	Transcript2
+     C0000570	128290	128405	Transcript2
+     C0000571	72845	73133	Transcript3
+     C0000571	73211	73274	Transcript3
+
+2. Coverage_file - coordinate sorted file in bedgraph format - it can contain coverage data (usually normalized) from RNAseq study, ChIPseq, ATACseq and so on.
+
+Coverage_file format example
+
+     C0000570	10481	10549	0.310587
+     C0000570	10579	10610	0.41057
+     C0000570	128288	128293	1.105
+
+## _USAGE_
+### Prerequisites (MUST be in your PATH)
+	BEDTOOLS
+	AWK
+    g++
+
+### Installation
+Download TRate
+
+cd TRate
+
+make		
+
+In file Trate.sh edit path to TRate folder, e.g.
+
+FOLDER_PATH="your/path/TRate"
+
+### Run TRate on test data
+
+     ./TRate.sh ./data/Exons_file.sbed ./data/Coverage_file.bg
+
+Output will be in a file Coverage_file.rate. 
+
+Output format
+
+     Transcript1	0.341895
+     Transcript2	1.98961
+     Transcript3	-1
+
+Transcript rate = -1 if no coverage data were found for this transcript.
+
+
+
+

TR.v1.cpp

@@ -0,0 +1,88 @@
+
+/* Changed command line processing interface compared to v2 */
+
+#include <unistd.h>
+#include <stdio.h>
+
+#include <fstream>
+#include <string>
+#include <string.h>
+#include <iostream>
+#include <sstream>
+#include <stdlib.h>
+
+using namespace std;
+
+int main( int argc , char** argv ) {
+
+
+   char fName[1000]="", fff[1000]="";
+
+   //printf ("optind=%d argv[optind]=%s strlen(argv[optind])=%d \n",optind,argv[optind],strlen(argv[optind]));
+    if (optind >= argc) {
+        fprintf(stderr, "Expected *.area.bed file\n");
+        fprintf(stderr, "Usage: *.area.bed\n");
+        exit(EXIT_FAILURE);
+    }
+
+   //strncat(fname,"",0);
+   strncat(fName,argv[optind],strlen(argv[optind]));
+   //printf ("fName=%s\n",fName);
+
+
+   std::ifstream area_file( fName ) ;
+   if (!area_file) {
+	fprintf(stderr, "Can't open input file %s\n",fName);	
+        exit(EXIT_FAILURE);
+    }
+
+   std::string line;
+   string trans="", trans_pred="", areas, widths;
+   int linecount=0;
+   float rate, area, width, area_sum=0, width_sum=0;
+
+   sprintf(fff, "%ste", fName);
+   std::ofstream outfile(fff);
+   printf("out=%s",fff);
+
+    while ( getline( area_file , line ) ) {
+	 //std::cout << linecount << ": " << line << '\n' ;//supposing '\n' to be line end
+	 //linecount++ ;
+	 stringstream ss(line);
+         getline(ss,trans, '\t');
+	 //	cout << "Transcript: " << trans << " trans_pred: "<<trans_pred<<endl;
+         getline(ss,areas, '\t');
+	 area=atof(areas.c_str());
+         //       cout << "area: " << area <<endl;
+         getline(ss,widths, '\t');
+         width=atof(widths.c_str());
+         //       cout << "width: " << width <<endl;
+
+        if(trans_pred != trans) 
+         { 
+          if(width_sum>0)  rate=area_sum/width_sum;
+          else rate=-1;
+
+          if(trans_pred!="") outfile << trans_pred << "\t" << rate << endl;   
+
+           area_sum=area;
+           width_sum=width;
+	 }
+        else
+           { area_sum+=area;
+             width_sum+=width;
+           }
+
+         trans_pred=trans;
+        }
+
+   //process last record
+    if(width_sum>0)  rate=area_sum/width_sum;
+          else rate=-1;
+
+          outfile << trans_pred << "\t" << rate << endl;   
+
+   outfile.close();
+   area_file.close();
+  cout << endl << "END";
+}

TRate.sh

@@ -0,0 +1,54 @@
+#!/bin/bash
+
+
+FOLDER_PATH="/home/newnataliya/From_timnatuk/APPS/APPS_NT/TRate"
+#Transcript_rate version 1
+
+# Bedtools should be installed in your Path!
+
+#This program takes in two arguments in fixed order: 
+#     Exons file - sorted by coordinates bed file with location of exons belonging to the transcript provided in column 4
+#format example
+# C0000568	109816	110194	AMEX60DDU001000001.1.bed
+# C0000570	39532	39624	AMEX60DDU001000002.1.bed
+# C0000570	39699	39934	AMEX60DDU001000002.1.bed
+#     Coverage file in sorted bedgraph format - it can contain coverage data from RNAseq study, ChIPseq, ATACseq and so on
+# format example
+#  C0000568	129865	129896	0.073524
+#  C0000568	129896	129965	0.036762
+#  C0000570	128344	128444	0.036762 
+# This program computes outputs "rate" of each transcript in exons file. Rate is computed as total mass of exons within the transcript divided by total length of these exons, mass is taken as approximation of the area under coverage curve, i.e. sum of areas of coverage rectangles.  
+
+#$1 EXONS_by_transcripts.sbed
+#$2 SRR2885267.sam.bam.norm0.036762.bg.s
+Exons_file=$1
+Cov_file=$2
+
+if [ ! -s "$Exons_file" ]; then
+    echo "Exons file $Exons_file is empty or doesn't exist, exit now"; 
+    exit;
+fi
+
+if [ ! -s "$Cov_file" ]; then
+    echo "Coverage file $Cov_file is empty or doesn't exist, exit now";
+    exit;
+fi
+
+if [ ! -s "EXONS_by_transcripts.tmp.sbed" ]; then
+   cp $Exons_file EXONS_by_transcripts.tmp.sbed;
+fi
+
+Name1=$(basename $Cov_file)
+#echo "Name1="$Name1
+
+#echo "Cov_file="$Cov_file
+
+bedtools intersect -sorted -wao -a EXONS_by_transcripts.tmp.sbed -b stdin < $Cov_file | awk '{print $4"\t"$8*$9"\t"$9}'  | sort -k1,1 > $Name1".ra"
+
+$FOLDER_PATH/TRate $Name1".ra" > $Name1".rate.log"
+
+echo "File $Cov_file processed, result is in "$Name1".rate"
+echo "Removing intermediate files";
+rm $Name1".ra"
+
+

data/Coverage_file.bg

@@ -0,0 +1,19 @@
+C0000570	10481	10549	0.310587
+C0000570	10579	10610	0.41057
+C0000570	128288	128293	1.105
+C0000570	128293	128294	2.031761
+C0000570	128294	128306	2.042348
+C0000570	128306	128316	2.352935
+C0000570	128316	128317	2.074109
+C0000570	128317	128322	2.084696
+C0000570	128322	128353	2.095283
+C0000570	128353	128388	2.10587
+C0000570	128388	128393	2.0983
+C0000570	128393	128394	2.074109
+C0000570	128394	128406	1.063522
+C0000570	128406	128416	1.052935
+C0000570	128416	128417	1.031761
+C0000570	128417	128420	1.021174
+C0000570	128420	128452	1.010587
+C0000571	33336	33436	5.021174
+C0000571	33437	33440	5.1234

data/Exons_file.sbed

@@ -0,0 +1,5 @@
+C0000570	10420	10640	Transcript1
+C0000570	128078	128167	Transcript2
+C0000570	128290	128405	Transcript2
+C0000571	72845	73133	Transcript3
+C0000571	73211	73274	Transcript3