From f82d2d225ac99287a5be11dd7f1e5447c43087fe Mon Sep 17 00:00:00 2001
From: Aaron <aaronquinlan@gmail.com>
Date: Thu, 17 Nov 2011 15:53:00 -0500
Subject: [PATCH] Added intervals option to tagBam.
---
src/tagBam/tagBam.cpp | 36 +++++++++++++++++++++++++++---------
src/tagBam/tagBam.h | 3 ++-
src/tagBam/tagBamMain.cpp | 30 ++++++++++++++++++++++++++----
3 files changed, 55 insertions(+), 14 deletions(-)
diff --git a/src/tagBam/tagBam.cpp b/src/tagBam/tagBam.cpp
index f8d1d7ea..d7474e04 100644
--- a/src/tagBam/tagBam.cpp
+++ b/src/tagBam/tagBam.cpp
@@ -15,7 +15,8 @@
// build
TagBam::TagBam(const string &bamFile, const vector<string> &annoFileNames,
const vector<string> &annoLables, const string &tag,
- bool useNames, bool useScores, bool sameStrand, bool diffStrand, float overlapFraction):
+ bool useNames, bool useScores, bool useIntervals,
+ bool sameStrand, bool diffStrand, float overlapFraction):
_bamFile(bamFile),
_annoFileNames(annoFileNames),
@@ -23,6 +24,7 @@ TagBam::TagBam(const string &bamFile, const vector<string> &annoFileNames,
_tag(tag),
_useNames(useNames),
_useScores(useScores),
+ _useIntervals(useIntervals),
_sameStrand(sameStrand),
_diffStrand(diffStrand),
_overlapFraction(overlapFraction)
@@ -94,7 +96,7 @@ void TagBam::Tag() {
// grab the current annotation file.
BedFile *anno = _annoFiles[i];
- if (!_useNames && !_useScores) {
+ if (!_useNames && !_useScores && !_useIntervals) {
// add the label for this annotation file to tag if there is overlap
if (anno->FindOneOrMoreOverlapsPerBin(a.chrom, a.start, a.end, a.strand, _sameStrand, _diffStrand, _overlapFraction))
{
@@ -102,7 +104,7 @@ void TagBam::Tag() {
}
}
// use the score field
- else if (!_useNames && _useScores) {
+ else if (!_useNames && _useScores && !_useIntervals) {
anno->FindOverlapsPerBin(a.chrom, a.start, a.end, a.strand, hits, _sameStrand, _diffStrand);
for (size_t i = 0; i < hits.size(); ++i) {
annotations << hits[i].score;
@@ -112,11 +114,27 @@ void TagBam::Tag() {
hits.clear();
}
// use the name field from the annotation files to populate tag
- else if (_useNames && !_useScores) {
+ else if (_useNames && !_useScores && !_useIntervals) {
anno->FindOverlapsPerBin(a.chrom, a.start, a.end, a.strand, hits, _sameStrand, _diffStrand);
- for (size_t i = 0; i < hits.size(); ++i) {
- annotations << hits[i].name;
- if (i < hits.size() - 1) annotations << ",";
+ for (size_t j = 0; j < hits.size(); ++i) {
+ annotations << hits[j].name;
+ if (j < hits.size() - 1) annotations << ",";
+ }
+ if (hits.size() > 0) annotations << ";";
+ hits.clear();
+ }
+ // use the name field from the annotation files to populate tag
+ else if (!_useNames && !_useScores && _useIntervals) {
+ anno->FindOverlapsPerBin(a.chrom, a.start, a.end, a.strand, hits, _sameStrand, _diffStrand);
+ for (size_t j = 0; j < hits.size(); ++j) {
+ annotations << _annoLabels[i] << ":" <<
+ hits[j].chrom << ":" <<
+ hits[j].start << "-" <<
+ hits[j].end << "," <<
+ hits[j].name << "," <<
+ hits[j].score << "," <<
+ hits[j].strand;
+ if (j < hits.size() - 1) annotations << ",";
}
if (hits.size() > 0) annotations << ";";
hits.clear();
@@ -126,11 +144,11 @@ void TagBam::Tag() {
if (annotations.str().size() > 0) {
al.AddTag(_tag, "Z", annotations.str().substr(0, annotations.str().size() - 1)); // get rid of the last ";"
}
- writer.SaveAlignment(al);
}
+ writer.SaveAlignment(al);
}
reader.Close();
-
+ writer.Close();
// close the annotations files;
CloseAnnoFiles();
}
diff --git a/src/tagBam/tagBam.h b/src/tagBam/tagBam.h
index 689dd0ac..082f38b7 100644
--- a/src/tagBam/tagBam.h
+++ b/src/tagBam/tagBam.h
@@ -41,7 +41,7 @@ public:
// constructor
TagBam(const string &bamFile, const vector<string> &annoFileNames,
const vector<string> &annoLabels, const string &tag,
- bool useNames, bool useScores, bool sameStrand,
+ bool useNames, bool useScores, bool useIntervals, bool sameStrand,
bool diffStrand, float overlapFraction);
// destructor
@@ -66,6 +66,7 @@ private:
// should we use the name field from the annotation files?
bool _useNames;
bool _useScores;
+ bool _useIntervals;
// do we care about strandedness when tagging?
bool _sameStrand;
diff --git a/src/tagBam/tagBamMain.cpp b/src/tagBam/tagBamMain.cpp
index 6bef5734..b4c750dc 100644
--- a/src/tagBam/tagBamMain.cpp
+++ b/src/tagBam/tagBamMain.cpp
@@ -38,6 +38,7 @@ int main(int argc, char* argv[]) {
bool haveFraction = false;
bool useNames = false;
bool useScores = false;
+ bool useIntervals = false;
bool sameStrand = false;
bool diffStrand = false;
bool haveBam = false;
@@ -109,6 +110,9 @@ int main(int argc, char* argv[]) {
else if (PARAMETER_CHECK("-scores", 7, parameterLength)) {
useScores = true;
}
+ else if (PARAMETER_CHECK("-intervals", 10, parameterLength)) {
+ useIntervals = true;
+ }
else if (PARAMETER_CHECK("-s", 2, parameterLength)) {
sameStrand = true;
}
@@ -156,13 +160,29 @@ int main(int argc, char* argv[]) {
cerr << endl << "*****" << endl << "*****ERROR: Use -scores or -names, not both. " << endl << "*****" << endl;
showHelp = true;
}
+ if (useScores && useIntervals) {
+ cerr << endl << "*****" << endl << "*****ERROR: Use -scores or -intervals, not both. " << endl << "*****" << endl;
+ showHelp = true;
+ }
+ if (useNames && useIntervals) {
+ cerr << endl << "*****" << endl << "*****ERROR: Use -names or -intervals, not both. " << endl << "*****" << endl;
+ showHelp = true;
+ }
+ if (!haveLabels && useIntervals) {
+ cerr << endl << "*****" << endl << "*****ERROR: Supply -labels when using -intervals. " << endl << "*****" << endl;
+ showHelp = true;
+ }
if (haveTag && tag.size() > 2) {
cerr << endl << "*****" << endl << "*****ERROR: Custom tags should be at most two characters per the SAM specification. " << endl << "*****" << endl;
showHelp = true;
}
+
if (!showHelp) {
- TagBam *ba = new TagBam(bamFile, inputFiles, inputLabels, tag, useNames, useScores, sameStrand, diffStrand, overlapFraction);
+ TagBam *ba = new TagBam(bamFile, inputFiles, inputLabels,
+ tag, useNames, useScores,
+ useIntervals, sameStrand, diffStrand,
+ overlapFraction);
ba->Tag();
delete ba;
return 0;
@@ -201,9 +221,11 @@ void ShowHelp(void) {
cerr << "\t-names\t" << "Use the name field from the annotation files to populate tags." << endl;
cerr << "\t\tBy default, the -labels values are used." << endl << endl;
- cerr << "\t-scores\t" << "A list of 1-based columns for each annotation file" << endl;
- cerr << "\t\tin which a color can be found." << endl << endl;
-
+ cerr << "\t-scores\t" << "Use the score field from the annotation files to populate tags." << endl;
+ cerr << "\t\tBy default, the -labels values are used." << endl << endl;
+
+ cerr << "\t-intervals\t" << "Use the full interval (including name, score, and strand) to populate tags." << endl;
+ cerr << "\t\t\tRequires the -labels option to identify from which file the interval came." << endl << endl;
exit(1);
}
--
GitLab