From da9026ba33206cd1a307a9e8639bf32722cb6b4f Mon Sep 17 00:00:00 2001
From: Brent Pedersen <bpederse@gmail.com>
Date: Mon, 29 Oct 2012 11:40:15 -0600
Subject: [PATCH] start of allowing to get concatenated sequence of exons
---
src/fastaFromBed/Makefile | 4 +++-
src/fastaFromBed/fastaFromBed.cpp | 19 +++++++++++++++++--
src/fastaFromBed/fastaFromBed.h | 4 +++-
src/fastaFromBed/fastaFromBedMain.cpp | 7 ++++++-
4 files changed, 29 insertions(+), 5 deletions(-)
diff --git a/src/fastaFromBed/Makefile b/src/fastaFromBed/Makefile
index 7b6d9823..dfbaeaa7 100644
--- a/src/fastaFromBed/Makefile
+++ b/src/fastaFromBed/Makefile
@@ -9,7 +9,9 @@ INCLUDES = -I$(UTILITIES_DIR)/bedFile/ \
-I$(UTILITIES_DIR)/gzstream/ \
-I$(UTILITIES_DIR)/genomeFile/ \
-I$(UTILITIES_DIR)/lineFileUtilities/ \
+ -I$(UTILITIES_DIR)/BamTools/include \
-I$(UTILITIES_DIR)/sequenceUtilities/ \
+ -I$(UTILITIES_DIR)/BlockedIntervals \
-I$(UTILITIES_DIR)/fileType/ \
-I$(UTILITIES_DIR)/Fasta/ \
-I$(UTILITIES_DIR)/version/
@@ -34,4 +36,4 @@ clean:
@echo "Cleaning up."
@rm -f $(OBJ_DIR)/fastaFromBedMain.o $(OBJ_DIR)/fastaFromBed.o
-.PHONY: clean
\ No newline at end of file
+.PHONY: clean
diff --git a/src/fastaFromBed/fastaFromBed.cpp b/src/fastaFromBed/fastaFromBed.cpp
index 1ad92a9f..399ac3dd 100644
--- a/src/fastaFromBed/fastaFromBed.cpp
+++ b/src/fastaFromBed/fastaFromBed.cpp
@@ -11,10 +11,11 @@
******************************************************************************/
#include "lineFileUtilities.h"
#include "fastaFromBed.h"
+#include "bedFile.h"
Bed2Fa::Bed2Fa(bool useName, const string &dbFile, const string &bedFile,
- const string &fastaOutFile, bool useFasta, bool useStrand) {
+ const string &fastaOutFile, bool useFasta, bool useStrand, bool useExons) {
_useName = useName;
_dbFile = dbFile;
@@ -22,6 +23,7 @@ Bed2Fa::Bed2Fa(bool useName, const string &dbFile, const string &bedFile,
_fastaOutFile = fastaOutFile;
_useFasta = useFasta;
_useStrand = useStrand;
+ _useExons = useExons;
_bed = new BedFile(_bedFile);
@@ -120,7 +122,20 @@ void Bed2Fa::ExtractDNA() {
if ( (bed.start <= seqLength) && (bed.end <= seqLength) )
{
int length = bed.end - bed.start;
- sequence = fr->getSubSequence(bed.chrom, bed.start, length);
+ if(_useExons){
+ bedVector bedBlocks; // vec to store the discrete BED "blocks"
+ GetBedBlocks(bed, bedBlocks);
+ sequence.clear();
+ for (int i = 0; i < (int) bedBlocks.size(); ++i) {
+ #cerr << bedBlocks[i].start << " " << bedBlocks[i].end << endl;
+ sequence += fr->getSubSequence(bed.chrom,
+ bedBlocks[i].start,
+ bedBlocks[i].end - bedBlocks[i].start);
+
+ }
+ } else {
+ sequence = fr->getSubSequence(bed.chrom, bed.start, length);
+ }
ReportDNA(bed, sequence);
}
else
diff --git a/src/fastaFromBed/fastaFromBed.h b/src/fastaFromBed/fastaFromBed.h
index 3f5f454b..0ab73118 100644
--- a/src/fastaFromBed/fastaFromBed.h
+++ b/src/fastaFromBed/fastaFromBed.h
@@ -13,6 +13,7 @@
#define FASTAFROMBED_H
#include "bedFile.h"
+#include "BlockedIntervals.h"
#include "sequenceUtils.h"
#include "Fasta.h"
#include <vector>
@@ -30,7 +31,7 @@ public:
// constructor
Bed2Fa(bool useName, const string &dbFile, const string &bedFile, const string &fastaOutFile,
- bool useFasta, bool useStrand);
+ bool useFasta, bool useStrand, bool useExons);
// destructor
~Bed2Fa(void);
@@ -47,6 +48,7 @@ private:
string _fastaOutFile;
bool _useFasta;
bool _useStrand;
+ bool _useExons;
// instance of a bed file class.
BedFile *_bed;
diff --git a/src/fastaFromBed/fastaFromBedMain.cpp b/src/fastaFromBed/fastaFromBedMain.cpp
index 9be9afb3..c5da8be2 100644
--- a/src/fastaFromBed/fastaFromBedMain.cpp
+++ b/src/fastaFromBed/fastaFromBedMain.cpp
@@ -43,6 +43,7 @@ int fastafrombed_main(int argc, char* argv[]) {
bool useNameOnly = false;
bool useFasta = true;
bool useStrand = false;
+ bool useExons = false;
// check to see if we should print out some help
if(argc <= 1) showHelp = true;
@@ -87,6 +88,9 @@ int fastafrombed_main(int argc, char* argv[]) {
else if(PARAMETER_CHECK("-name", 5, parameterLength)) {
useNameOnly = true;
}
+ else if(PARAMETER_CHECK("-exons", 6, parameterLength)) {
+ useExons = true;
+ }
else if(PARAMETER_CHECK("-tab", 4, parameterLength)) {
useFasta = false;
}
@@ -105,7 +109,7 @@ int fastafrombed_main(int argc, char* argv[]) {
if (!showHelp) {
- Bed2Fa *b2f = new Bed2Fa(useNameOnly, fastaDbFile, bedFile, fastaOutFile, useFasta, useStrand);
+ Bed2Fa *b2f = new Bed2Fa(useNameOnly, fastaDbFile, bedFile, fastaOutFile, useFasta, useStrand, useExons);
delete b2f;
}
else {
@@ -127,6 +131,7 @@ void fastafrombed_help(void) {
cerr << "\t-bed\tBED/GFF/VCF file of ranges to extract from -fi" << endl;
cerr << "\t-fo\tOutput file (can be FASTA or TAB-delimited)" << endl;
cerr << "\t-name\tUse the name field for the FASTA header" << endl;
+ cerr << "\t-exons\tgiven a bed12, concat the sequence from exons" << endl;
cerr << "\t-tab\tWrite output in TAB delimited format." << endl;
cerr << "\t\t- Default is FASTA format." << endl << endl;
--
GitLab