/**
* Determine if the given loc overlaps any loc in the sorted set
*
* @param loc the location to test
* @return
*/
public boolean overlaps(final GenomeLoc loc) {
for (final GenomeLoc e : mArray) {
if (e.overlapsP(loc)) {
return true;
}
}
return false;
}
public GenomeLocSortedSet subtractRegions(GenomeLocSortedSet toRemoveSet) {
LinkedList<GenomeLoc> good = new LinkedList<GenomeLoc>();
Stack<GenomeLoc> toProcess = new Stack<GenomeLoc>();
Stack<GenomeLoc> toExclude = new Stack<GenomeLoc>();
// initialize the stacks
toProcess.addAll(mArray);
Collections.reverse(toProcess);
toExclude.addAll(toRemoveSet.mArray);
Collections.reverse(toExclude);
int i = 0;
while (!toProcess.empty()) { // while there's still stuff to process
if (toExclude.empty()) {
good.addAll(toProcess); // no more excludes, all the processing stuff is good
break;
}
GenomeLoc p = toProcess.peek();
GenomeLoc e = toExclude.peek();
if (p.overlapsP(e)) {
toProcess.pop();
for (GenomeLoc newP : p.subtract(e)) toProcess.push(newP);
} else if (p.compareContigs(e) < 0) {
good.add(toProcess.pop()); // p is now good
} else if (p.compareContigs(e) > 0) {
toExclude.pop(); // e can't effect anything
} else if (p.getStop() < e.getStart()) {
good.add(toProcess.pop()); // p stops before e starts, p is good
} else if (e.getStop() < p.getStart()) {
toExclude.pop(); // p starts after e stops, e is done
} else {
throw new ReviewedStingException("BUG: unexpected condition: p=" + p + ", e=" + e);
}
if (i++ % 10000 == 0) logger.debug("removeRegions operation: i = " + i);
}
return createSetFromList(genomeLocParser, good);
}
/**
* Fully qualified constructor: instantiates a new GATKFeatureRecordList object with specified
* GATKFeature track name, location on the reference, and list of associated GATKFeatures. This is
* a knee-deep COPY constructor: passed name, loc, and data element objects will be referenced
* from the created GATKFeatureRecordList (so that changing them from outside will affect data in
* this object), however, the data elements will be copied into a newly allocated list, so that
* the 'data' collection argument can be modified afterwards without affecting the state of this
* record list. WARNING: this constructor is (semi-)validating: passed name and location are
* allowed to be nulls (although it maybe unsafe, use caution), but if they are not nulls, then
* passed non-null GATKFeature data elements must have same track name, and their locations must
* overlap with the passed 'location' argument. Null data elements or null 'data' collection
* argument are allowed as well.
*
* @param name the name of the track
* @param data the collection of features at this location
* @param loc the location
*/
public RODRecordListImpl(String name, Collection<GATKFeature> data, GenomeLoc loc) {
this.records = new ArrayList<GATKFeature>(data == null ? 0 : data.size());
this.name = name;
this.location = loc;
if (data == null || data.size() == 0) return; // empty dataset, nothing to do
for (GATKFeature r : data) {
records.add(r);
if (r == null) continue;
if (!this.name.equals(r.getName())) {
throw new ReviewedStingException(
"Attempt to add GATKFeature with non-matching name "
+ r.getName()
+ " to the track "
+ name);
}
if (location != null && !location.overlapsP(r.getLocation())) {
throw new ReviewedStingException(
"Attempt to add GATKFeature that lies outside of specified interval "
+ location
+ "; offending GATKFeature:\n"
+ r.toString());
}
}
}