From 5f48036660c1ac96da1e7bac0cfe2680caa374f7 Mon Sep 17 00:00:00 2001
From: Aaron <aaronquinlan@gmail.com>
Date: Fri, 16 Sep 2011 10:04:39 -0400
Subject: [PATCH] chromsweep is a proper tool. Now to make it an iterable
class.
---
src/chromsweep/chromsweep.cpp | 69 +++++++------------------------
src/chromsweep/chromsweep.h | 17 +-------
src/chromsweep/chromsweepMain.cpp | 8 +---
3 files changed, 17 insertions(+), 77 deletions(-)
diff --git a/src/chromsweep/chromsweep.cpp b/src/chromsweep/chromsweep.cpp
index 2211c9ff..e36dcea4 100644
--- a/src/chromsweep/chromsweep.cpp
+++ b/src/chromsweep/chromsweep.cpp
@@ -55,6 +55,10 @@ ChromSweep::ChromSweep(string bedAFile, string bedBFile, bool anyHit,
// prime the results pump.
_qy_lineNum = 0;
_db_lineNum = 0;
+
+ _hits.reserve(1000);
+ _cache.reserve(1000);
+
_bedA->Open();
_bedB->Open();
_qy_status = _bedA->GetNextBed(_curr_qy, _qy_lineNum);
@@ -125,72 +129,29 @@ void ChromSweep::ChromCheck()
}
else if (_curr_qy.chrom < _curr_db.chrom) {
// report hits for the remaining queries on this chrom
- BED tmp_curr_qy = _curr_qy;
- while (!_bedA->Empty() && tmp_curr_qy.chrom == _curr_qy.chrom)
+ string curr_chrom = _curr_qy.chrom;
+ while (!_bedA->Empty() && _curr_qy.chrom == curr_chrom)
{
ScanCache();
- _results.push(make_pair(tmp_curr_qy, _hits));
- _qy_status = _bedA->GetNextBed(tmp_curr_qy, _qy_lineNum);
+ _results.push(make_pair(_curr_qy, _hits));
+ _qy_status = _bedA->GetNextBed(_curr_qy, _qy_lineNum);
_hits.clear();
}
// now fast forward query to catch up to database
- while (!_bedA->Empty() && tmp_curr_qy.chrom < _curr_db.chrom)
+ while (!_bedA->Empty() && _curr_qy.chrom < _curr_db.chrom)
{
// hits is empty to reflect the fact that no hits are found in catch-up mode
- _results.push(make_pair(tmp_curr_qy, _hits));
- _qy_status = _bedA->GetNextBed(tmp_curr_qy, _qy_lineNum);
+ _results.push(make_pair(_curr_qy, _hits));
+ _qy_status = _bedA->GetNextBed(_curr_qy, _qy_lineNum);
}
- _curr_qy = tmp_curr_qy;
_cache.clear();
}
}
-//
-// void ChromSweep::ChromCheck(BED &curr_qy, BED &curr_db,
-// BedLineStatus &qy_status, BedLineStatus &db_status,
-// int &qy_lineNum, int &db_lineNum,
-// vector<BED> &db_cache, vector<BED> &hits)
-// {
-// if ((curr_qy.chrom == curr_db.chrom) || (db_status == BED_INVALID) || (qy_status == BED_INVALID)) {
-// return;
-// }
-//
-// if (curr_qy.chrom > curr_db.chrom) {
-// while (!_bedB->Empty() && curr_db.chrom < curr_qy.chrom)
-// {
-// db_status = _bedB->GetNextBed(curr_db, db_lineNum);
-// }
-// db_cache.clear();
-// }
-// else if (curr_qy.chrom < curr_db.chrom) {
-// // report hits for the remaining queries on this chrom
-// BED tmp_curr_qy = curr_qy;
-// while (!_bedA->Empty() && tmp_curr_qy.chrom == curr_qy.chrom)
-// {
-// //db_cache = ScanCache(tmp_curr_qy, qy_status, db_cache, hits);
-// ScanCache(tmp_curr_qy, qy_status, db_cache, hits);
-//
-// //ReportHits(tmp_curr_qy, hits);
-// _results.push(make_pair(tmp_curr_qy, hits));
-// qy_status = _bedA->GetNextBed(tmp_curr_qy, qy_lineNum);
-// hits.clear();
-// }
-// // now fast forward query to catch up to database
-// while (!_bedA->Empty() && tmp_curr_qy.chrom < curr_db.chrom)
-// {
-// // hits is empty to reflect the fact that no hits are found in catch-up mode
-// ReportHits(tmp_curr_qy, hits);
-// qy_status = _bedA->GetNextBed(tmp_curr_qy, qy_lineNum);
-// }
-// curr_qy = tmp_curr_qy;
-// db_cache.clear();
-// }
-// }
-void ChromSweep::ReportHits(const BED &curr_qy, const vector<BED> &hits) {
- _bedA->reportBedTab(curr_qy);
- cout << hits.size() << endl;
+void ChromSweep::ReportQuery(const BED &query) {
+ _bedA->reportBedTab(query);
}
@@ -201,9 +162,7 @@ bool ChromSweep::Next(pair<BED, vector<BED> > &next) {
// scan the database cache for hits
ScanCache();
// advance the db until we are ahead of the query. update hits and cache as necessary
- while (!_bedB->Empty() &&
- _curr_qy.chrom == _curr_db.chrom &&
- !(after(_curr_db, _curr_qy)))
+ while (!_bedB->Empty() && _curr_qy.chrom == _curr_db.chrom && !(after(_curr_db, _curr_qy)))
{
if (overlaps(_curr_qy.start, _curr_qy.end, _curr_db.start, _curr_db.end) > 0) {
_hits.push_back(_curr_db);
diff --git a/src/chromsweep/chromsweep.h b/src/chromsweep/chromsweep.h
index bbbe6bdc..1fb58405 100644
--- a/src/chromsweep/chromsweep.h
+++ b/src/chromsweep/chromsweep.h
@@ -44,7 +44,7 @@ public:
bool Next(pair<BED, vector<BED> > &next);
- //pair<BED, vector<BED> > GetNextResult(void);
+ void ReportQuery(const BED &query);
private:
@@ -90,23 +90,8 @@ private:
BedLineStatus _qy_status, _db_status;
int _qy_lineNum, _db_lineNum;
- //------------------------------------------------
- // private methods
- //------------------------------------------------
- // void ScanCache(const BED &curr_qy, BedLineStatus qy_status, vector<BED> &db_cache, vector<BED> &hits);
- //
- // void ChromCheck(BED &curr_qy, BED &curr_db,
- // BedLineStatus &qy_status, BedLineStatus &db_status,
- // int &qy_lineNum, int &db_lineNum,
- // vector<BED> &db_cache, vector<BED> &hits);
-
void ScanCache();
void ChromCheck();
-
- void Sweep();
- void ReportHits(const BED &curr_qy, const vector<BED> &hits);
-
-
};
#endif /* CHROMSWEEP_H */
diff --git a/src/chromsweep/chromsweepMain.cpp b/src/chromsweep/chromsweepMain.cpp
index aebed1a7..90ed322c 100644
--- a/src/chromsweep/chromsweepMain.cpp
+++ b/src/chromsweep/chromsweepMain.cpp
@@ -201,13 +201,9 @@ int main(int argc, char* argv[]) {
pair<BED, vector<BED> > hit_set;
while (sweep->Next(hit_set)) {
- cout << hit_set.first.chrom << "\t"
- << hit_set.first.start << "\t"
- << hit_set.first.end << "\t"
- << hit_set.second.size() << "\n";
+ sweep->ReportQuery(hit_set.first);
+ cout << hit_set.second.size() << "\n";
}
-
-
delete sweep;
return 0;
}
--
GitLab