diff --git a/bin/closestBed b/bin/closestBed
index 32eabe225ce040d0039862be7e70573b1911c189..7d70fa4c9c24981be21fa2db2743e4b3d8fdc37a 100755
Binary files a/bin/closestBed and b/bin/closestBed differ
diff --git a/bin/intersectBed b/bin/intersectBed
index daa7c51268dea86631638952fabb00c43d7f79cd..d1728c9b3a0bc452860325bf8f93bdeab2ccf251 100755
Binary files a/bin/intersectBed and b/bin/intersectBed differ
diff --git a/obj/closestBed.o b/obj/closestBed.o
index ff1fb98b41bfdcd2c32932a672a6c174912c0ceb..6ed2759ffc860f57d3d247fd5aad9cd699521061 100644
Binary files a/obj/closestBed.o and b/obj/closestBed.o differ
diff --git a/obj/closestMain.o b/obj/closestMain.o
index c1f70ff628ca8b38f56f1830309f1114de46adaf..149f4e12e9000ffb85b92e1f27d1e56373009284 100644
Binary files a/obj/closestMain.o and b/obj/closestMain.o differ
diff --git a/obj/intersectBed.o b/obj/intersectBed.o
index 2d4b72b244825c2465a9a69641fb8ab226bee7f0..aaecf35d53e628f6bc125e363cbd2b1ca8e2ee87 100644
Binary files a/obj/intersectBed.o and b/obj/intersectBed.o differ
diff --git a/obj/intersectMain.o b/obj/intersectMain.o
index 04c7b914c474a327935be62c3c645aab7aadf8ef..77731f5f8f7989950a3d5202a0d1c62706ee1f02 100644
Binary files a/obj/intersectMain.o and b/obj/intersectMain.o differ
diff --git a/src/closestBed/a.bed b/src/closestBed/a.bed
new file mode 100644
index 0000000000000000000000000000000000000000..5b1e447c6ff93e0490cfb9bc26d761a0677b9714
--- /dev/null
+++ b/src/closestBed/a.bed
@@ -0,0 +1,3 @@
+chr1 100 101 rs1234
+
+
diff --git a/src/closestBed/b.bed b/src/closestBed/b.bed
new file mode 100644
index 0000000000000000000000000000000000000000..807c0cd01f2c9c9841060c2adca918f19ab7c102
--- /dev/null
+++ b/src/closestBed/b.bed
@@ -0,0 +1,3 @@
+chr1 20000 30000 geneA 100 +
+chr1 20000 30000 geneB 100 -
+
diff --git a/src/closestBed/closestBed.cpp b/src/closestBed/closestBed.cpp
index 0bf1326753b0f1b8aa34de8775ba81c9bdc0a498..05c7657afe63f9a8437ec959accf2f03fec64738 100755
--- a/src/closestBed/closestBed.cpp
+++ b/src/closestBed/closestBed.cpp
@@ -1,12 +1,14 @@
-//
-// closestBed.cpp
-// BEDTools
-//
-// Created by Aaron Quinlan Spring 2009.
-// Copyright 2009 Aaron Quinlan. All rights reserved.
-//
-// Summary: Looks for the closest features in two BED files.
-//
+/*****************************************************************************
+ closestBed.cpp
+
+ (c) 2009 - Aaron Quinlan
+ Hall Laboratory
+ Department of Biochemistry and Molecular Genetics
+ University of Virginia
+ aaronquinlan@gmail.com
+
+ Licenced under the GNU General Public License 2.0+ license.
+******************************************************************************/
#include "lineFileUtilities.h"
#include "closestBed.h"
@@ -188,66 +190,53 @@ void BedClosest::FindWindowOverlaps(BED &a, vector<BED> &hits) {
}
-void BedClosest::ClosestBed() {
+void BedClosest::ClosestBed(istream &bedInput) {
// load the "B" bed file into a map so
// that we can easily compare "A" to it for overlaps
bedB->loadBedFileIntoMap();
-
- string bedLine;
- BED bedEntry;
- int lineNum = 0;
-
- // open the BED file for reading
- if (bedA->bedFile != "stdin") {
-
- ifstream bed(bedA->bedFile.c_str(), ios::in);
- if ( !bed ) {
- cerr << "Error: The requested bed file (" <<bedA->bedFile << ") could not be opened. Exiting!" << endl;
- exit (1);
- }
+ string bedLine;
+ int lineNum = 0; // current input line number
+ vector<BED> hits; // vector of potential hits
+ vector<string> bedFields; // vector for a BED entry
+
+ // reserve some space
+ hits.reserve(100);
+ bedFields.reserve(12);
- BED a;
- while (getline(bed, bedLine)) {
-
- if ((bedLine.find("track") != string::npos) || (bedLine.find("browser") != string::npos)) {
- continue;
- }
- else {
- vector<string> bedFields;
- Tokenize(bedLine,bedFields);
+ BED a;
+ // process each entry in A
+ while (getline(bedInput, bedLine)) {
- lineNum++;
- if (bedA->parseBedLine(a, bedFields, lineNum)) {
- vector<BED> hits;
- FindWindowOverlaps(a, hits);
- }
- }
+ lineNum++;
+ Tokenize(bedLine,bedFields);
+
+ // find the overlaps with B if it's a valid BED entry.
+ if (bedA->parseLine(a, bedFields, lineNum)) {
+ FindWindowOverlaps(a, hits);
+ hits.clear();
}
- }
- // "A" is being passed via STDIN.
- else {
- BED a;
- while (getline(cin, bedLine)) {
+ // reset for the next input line
+ bedFields.clear();
+ }
+}
+// END ClosestBed
- if ((bedLine.find("track") != string::npos) || (bedLine.find("browser") != string::npos)) {
- continue;
- }
- else {
- vector<string> bedFields;
- Tokenize(bedLine,bedFields);
- lineNum++;
- if (bedA->parseBedLine(a, bedFields, lineNum)) {
- vector<BED> hits;
- FindWindowOverlaps(a, hits);
- }
- }
+
+void BedClosest::DetermineBedInput() {
+ if (bedA->bedFile != "stdin") { // process a file
+ ifstream beds(bedA->bedFile.c_str(), ios::in);
+ if ( !beds ) {
+ cerr << "Error: The requested bed file (" << bedA->bedFile << ") could not be opened. Exiting!" << endl;
+ exit (1);
}
+ ClosestBed(beds);
+ }
+ else { // process stdin
+ ClosestBed(cin);
}
}
-// END ClosestBed
-
diff --git a/src/closestBed/closestBed.h b/src/closestBed/closestBed.h
index 720406d09f2fd032bbe0f93f1fdcf3a44147c11f..0e3804bc49956f32f3b42a6db7cc2534dc7b4046 100755
--- a/src/closestBed/closestBed.h
+++ b/src/closestBed/closestBed.h
@@ -1,3 +1,14 @@
+/*****************************************************************************
+ closestBed.h
+
+ (c) 2009 - Aaron Quinlan
+ Hall Laboratory
+ Department of Biochemistry and Molecular Genetics
+ University of Virginia
+ aaronquinlan@gmail.com
+
+ Licenced under the GNU General Public License 2.0+ license.
+******************************************************************************/
#ifndef CLOSESTBED_H
#define CLOSESTBED_H
@@ -25,8 +36,9 @@ public:
void reportB(const BED &);
void reportNullB();
- void ClosestBed();
+ void ClosestBed(istream &bedInput);
void FindWindowOverlaps(BED &, vector<BED> &);
+ void DetermineBedInput();
private:
diff --git a/src/closestBed/closestMain.cpp b/src/closestBed/closestMain.cpp
index 7ec0c26949240187f0b04b4714e602fa826c93fa..bb30410d2ba62f6506d4e68a567d1eeed32efb1c 100755
--- a/src/closestBed/closestMain.cpp
+++ b/src/closestBed/closestMain.cpp
@@ -1,3 +1,14 @@
+/*****************************************************************************
+ closestMain.cpp
+
+ (c) 2009 - Aaron Quinlan
+ Hall Laboratory
+ Department of Biochemistry and Molecular Genetics
+ University of Virginia
+ aaronquinlan@gmail.com
+
+ Licenced under the GNU General Public License 2.0+ license.
+******************************************************************************/
#include "closestBed.h"
#include "version.h"
@@ -81,7 +92,7 @@ int main(int argc, char* argv[]) {
if (!showHelp) {
BedClosest *bc = new BedClosest(bedAFile, bedBFile, forceStrand, tieMode);
- bc->ClosestBed();
+ bc->DetermineBedInput();
return 0;
}
else {
@@ -96,24 +107,29 @@ void ShowHelp(void) {
cerr << " Aaron Quinlan, Ph.D. (aaronquinlan@gmail.com) " << endl ;
cerr << " Hall Laboratory, University of Virginia" << endl;
cerr << "===============================================" << endl << endl;
- cerr << "Description: For each feature in BED A, finds the closest feature (upstream or downstream) in BED B" << endl;
+ cerr << "SUMMARY: For each feature in BED A, finds the closest " << endl;
+ cerr << " feature (upstream or downstream) in BED B." << endl << endl;
- cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <a.bed> -b <b.bed>" << endl << endl;
+ cerr << "USAGE: " << PROGRAM_NAME << " [OPTIONS] -a <a.bed> -b <b.bed>" << endl << endl;
cerr << "OPTIONS: " << endl;
- cerr << "\t" << "-s\t\t\t" << "Force strandedness. Only report hits in B that overlap A on the same strand." << endl << "\t\t\t\tBy default, overlaps are reported without respect to strand." << endl << endl;
- cerr << "\t" << "-t\t\t\t" << "How to handle ties for closest features. This occurs when two features in B have exactly the" << endl;
- cerr << "\t\t\t\t" << "same overlap with a feature in A. By default, all such features in B are reported. Here are all the options:" << endl;
- cerr << "\t\t\t\t\t" << "all\t\tReport all ties (default)." << endl;
- cerr << "\t\t\t\t\t" << "first\t\tReport the first tie that occurred in the B file." << endl;
- cerr << "\t\t\t\t\t" << "last\t\tReport the last tie that occurred in the B file." << endl;
+ cerr << " " << "-s\t" << "Force strandedness. That is, find the closest feature in B that" << endl;
+ cerr << "\toverlaps A on the same strand." << endl;
+ cerr << "\t - By default, overlaps are reported without respect to strand." << endl << endl;
+
+ cerr << " " << "-t\t" << "How ties for closest feature are handled. This occurs when two" << endl;
+ cerr << "\tfeatures in B have exactly the same overlap with a feature in A." << endl;
+ cerr << "\tBy default, all such features in B are reported." << endl;
+ cerr << "\tHere are all the options:" << endl;
+ cerr << "\t - \"all\" Report all ties (default)." << endl;
+ cerr << "\t - \"first\" Report the first tie that occurred in the B file." << endl;
+ cerr << "\t - \"last\" Report the last tie that occurred in the B file." << endl << endl;
cerr << "NOTES: " << endl;
- cerr << "\t" << "-i stdin\t\t" << "Allows BED file A to be read from stdin. E.g.: cat a.bed | closestBed -a stdin -b B.bed" << endl << endl;
- cerr << "\t" << "Reports \"none\" for chrom and \"-1\" for all other fields when a feature is not found in B on the same chromosome" << endl;
- cerr << "\t" << "as the feature in A. E.g. none\t-1\t-1" << endl << endl;
- cerr << "\t***Only tab-delimited BED3 - BED6 formats allowed.***" << endl << endl;
+ cerr << " Reports \"none\" for chrom and \"-1\" for all other fields when a feature" << endl;
+ cerr << " is not found in B on the same chromosome as the feature in A." << endl;
+ cerr << " E.g. none\t-1\t-1" << endl << endl;
// end the program here
exit(1);
diff --git a/src/intersectBed/intersectMain.cpp b/src/intersectBed/intersectMain.cpp
index c4fdd9f2845ab3e27589abc381a87022f3f2179a..e0c657dbcd5ed366e389bd82f59b53c35289a583 100755
--- a/src/intersectBed/intersectMain.cpp
+++ b/src/intersectBed/intersectMain.cpp
@@ -156,11 +156,11 @@ void ShowHelp(void) {
cerr << " Aaron Quinlan, Ph.D. (aaronquinlan@gmail.com) " << endl ;
cerr << " Hall Laboratory, University of Virginia" << endl;
cerr << "===============================================" << endl << endl;
- cerr << "Description: Report overlaps between a.bed and b.bed." << endl << endl;
+ cerr << "SUMMARY: Report overlaps between a.bed and b.bed." << endl << endl;
- cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <a.bed> -b <b.bed>" << endl << endl;
+ cerr << "USAGE: " << PROGRAM_NAME << " [OPTIONS] -a <a.bed> -b <b.bed>" << endl << endl;
- cerr << "Options: " << endl;
+ cerr << "OPTIONS: " << endl;
cerr << " " << "-wa\t" << "Write the original entry in A for each overlap." << endl << endl;
cerr << " " << "-wb\t" << "Write the original entry in B for each overlap." << endl;
cerr << "\t - Useful for knowing _what_ A overlaps. Restricted by -f." << endl << endl;
diff --git a/src/windowBed/windowMain.cpp b/src/windowBed/windowMain.cpp
index d24289993aaaadf5771da4714fffd070c138704d..4e5a7502fba5321c00b5c0c6ff47cff866b70fc8 100755
--- a/src/windowBed/windowMain.cpp
+++ b/src/windowBed/windowMain.cpp
@@ -167,11 +167,11 @@ void ShowHelp(void) {
cerr << " Aaron Quinlan, Ph.D. (aaronquinlan@gmail.com) " << endl ;
cerr << " Hall Laboratory, University of Virginia" << endl;
cerr << "===============================================" << endl << endl;
- cerr << "Description: Examines a \"window\" around each feature in A and" << endl;
- cerr << "reports all features in B that overlap the window. For each" << endl;
- cerr << "overlap the entire entry in A and B are reported." << endl << endl;
+ cerr << "SUMMARY: Examines a \"window\" around each feature in A and" << endl;
+ cerr << " reports all features in B that overlap the window. For each" << endl;
+ cerr << " overlap the entire entry in A and B are reported." << endl << endl;
- cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <a.bed> -b <b.bed>" << endl << endl;
+ cerr << "USAGE: " << PROGRAM_NAME << " [OPTIONS] -a <a.bed> -b <b.bed>" << endl << endl;
cerr << "OPTIONS: " << endl;
cerr << " " << "-w\t" << "Base pairs added upstream and downstream of each entry" << endl;