@Test(expectedExceptions = IllegalArgumentException.class)
public void testAddDuplicate() {
assertTrue(mSortedSet.size() == 0);
GenomeLoc g = genomeLocParser.createGenomeLoc(contigOneName, 0, 0);
mSortedSet.add(g);
assertTrue(mSortedSet.size() == 1);
mSortedSet.add(g);
}
@Test(expectedExceptions = IllegalArgumentException.class)
public void addThrowsException() {
assertTrue(mSortedSet.size() == 0);
GenomeLoc g = genomeLocParser.createGenomeLoc(contigOneName, 1, 50);
mSortedSet.add(g);
GenomeLoc f = genomeLocParser.createGenomeLoc(contigOneName, 30, 80);
mSortedSet.add(f);
}
@Test
public void testAdd() {
GenomeLoc g = genomeLocParser.createGenomeLoc(contigOneName, 0, 0);
assertTrue(mSortedSet.size() == 0);
mSortedSet.add(g);
assertTrue(mSortedSet.size() == 1);
}
@Test
public void substractComplexExample() {
GenomeLoc e = genomeLocParser.createGenomeLoc(contigOneName, 1, 20);
mSortedSet.add(e);
GenomeLoc r1 = genomeLocParser.createGenomeLoc(contigOneName, 3, 5);
GenomeLoc r2 = genomeLocParser.createGenomeLoc(contigOneName, 10, 12);
GenomeLoc r3 = genomeLocParser.createGenomeLoc(contigOneName, 16, 18);
GenomeLocSortedSet toExclude =
new GenomeLocSortedSet(genomeLocParser, Arrays.asList(r1, r2, r3));
GenomeLocSortedSet remaining = mSortedSet.subtractRegions(toExclude);
// logger.debug("Initial " + mSortedSet);
// logger.debug("Exclude " + toExclude);
// logger.debug("Remaining " + remaining);
assertEquals(mSortedSet.coveredSize(), 20);
assertEquals(toExclude.coveredSize(), 9);
assertEquals(remaining.coveredSize(), 11);
Iterator<GenomeLoc> it = remaining.iterator();
GenomeLoc p1 = it.next();
GenomeLoc p2 = it.next();
GenomeLoc p3 = it.next();
GenomeLoc p4 = it.next();
assertEquals(genomeLocParser.createGenomeLoc(contigOneName, 1, 2), p1);
assertEquals(genomeLocParser.createGenomeLoc(contigOneName, 6, 9), p2);
assertEquals(genomeLocParser.createGenomeLoc(contigOneName, 13, 15), p3);
assertEquals(genomeLocParser.createGenomeLoc(contigOneName, 19, 20), p4);
}
@Test
public void overlap() {
for (int i = 1; i < 6; i++) {
final int start = i * 10;
mSortedSet.add(genomeLocParser.createGenomeLoc(contigOneName, start, start + 1));
}
// test matches in and around interval
assertFalse(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 9, 9)));
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 10, 10)));
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 11, 11)));
assertFalse(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 12, 12)));
// test matches spanning intervals
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 14, 20)));
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 11, 15)));
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 30, 40)));
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 51, 53)));
// test miss
assertFalse(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 12, 19)));
// test exact match after miss
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 40, 41)));
// test matches at beginning of intervals
assertFalse(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 5, 6)));
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 0, 10)));
// test matches at end of intervals
assertFalse(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 52, 53)));
assertTrue(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 51, 53)));
assertFalse(mSortedSet.overlaps(genomeLocParser.createGenomeLoc(contigOneName, 52, 53)));
}
/**
* create a list of genomic locations, given a reference sequence
*
* @param dict the sequence dictionary to create a collection from
* @return the GenomeLocSet of all references sequences as GenomeLoc's
*/
public static GenomeLocSortedSet createSetFromSequenceDictionary(SAMSequenceDictionary dict) {
GenomeLocParser parser = new GenomeLocParser(dict);
GenomeLocSortedSet returnSortedSet = new GenomeLocSortedSet(parser);
for (SAMSequenceRecord record : dict.getSequences()) {
returnSortedSet.add(
parser.createGenomeLoc(record.getSequenceName(), 1, record.getSequenceLength()));
}
return returnSortedSet;
}
@Test
public void addRegion() {
assertTrue(mSortedSet.size() == 0);
GenomeLoc g = genomeLocParser.createGenomeLoc(contigOneName, 1, 50);
mSortedSet.add(g);
GenomeLoc f = genomeLocParser.createGenomeLoc(contigOneName, 30, 80);
mSortedSet.addRegion(f);
assertTrue(mSortedSet.size() == 1);
}
@Test
public void deleteAllByRegion() {
GenomeLoc e = genomeLocParser.createGenomeLoc(contigOneName, 1, 100);
mSortedSet.add(e);
for (int x = 1; x < 101; x++) {
GenomeLoc del = genomeLocParser.createGenomeLoc(contigOneName, x, x);
mSortedSet = mSortedSet.subtractRegions(new GenomeLocSortedSet(genomeLocParser, del));
}
assertTrue(mSortedSet.isEmpty());
}
@Test
public void addRegionsOutOfOrder() {
final String contigTwoName = header.getSequenceDictionary().getSequence(2).getSequenceName();
assertTrue(mSortedSet.size() == 0);
GenomeLoc g = genomeLocParser.createGenomeLoc(contigTwoName, 1, 50);
mSortedSet.add(g);
GenomeLoc f = genomeLocParser.createGenomeLoc(contigOneName, 30, 80);
mSortedSet.addRegion(f);
assertTrue(mSortedSet.size() == 2);
assertTrue(mSortedSet.toList().get(0).getContig().equals(contigOneName));
assertTrue(mSortedSet.toList().get(1).getContig().equals(contigTwoName));
}
@Test
public void deleteSomeByRegion() {
GenomeLoc e = genomeLocParser.createGenomeLoc(contigOneName, 1, 100);
mSortedSet.add(e);
for (int x = 1; x < 50; x++) {
GenomeLoc del = genomeLocParser.createGenomeLoc(contigOneName, x, x);
mSortedSet = mSortedSet.subtractRegions(new GenomeLocSortedSet(genomeLocParser, del));
}
assertTrue(!mSortedSet.isEmpty());
assertTrue(mSortedSet.size() == 1);
GenomeLoc loc = mSortedSet.iterator().next();
assertTrue(loc.getStop() == 100);
assertTrue(loc.getStart() == 50);
}
@Test
public void mergingOverlappingAbove() {
GenomeLoc e = genomeLocParser.createGenomeLoc(contigOneName, 0, 50);
GenomeLoc g = genomeLocParser.createGenomeLoc(contigOneName, 49, 100);
assertTrue(mSortedSet.size() == 0);
mSortedSet.add(g);
assertTrue(mSortedSet.size() == 1);
mSortedSet.addRegion(e);
assertTrue(mSortedSet.size() == 1);
Iterator<GenomeLoc> iter = mSortedSet.iterator();
GenomeLoc loc = iter.next();
assertEquals(loc.getStart(), 0);
assertEquals(loc.getStop(), 100);
assertEquals(loc.getContigIndex(), 1);
}
@Test
public void deleteSuperRegion() {
GenomeLoc e = genomeLocParser.createGenomeLoc(contigOneName, 10, 20);
GenomeLoc g = genomeLocParser.createGenomeLoc(contigOneName, 70, 100);
mSortedSet.add(g);
mSortedSet.addRegion(e);
assertTrue(mSortedSet.size() == 2);
// now delete a region
GenomeLoc d = genomeLocParser.createGenomeLoc(contigOneName, 15, 75);
mSortedSet = mSortedSet.subtractRegions(new GenomeLocSortedSet(genomeLocParser, d));
Iterator<GenomeLoc> iter = mSortedSet.iterator();
GenomeLoc loc = iter.next();
assertTrue(loc.getStart() == 10);
assertTrue(loc.getStop() == 14);
assertTrue(loc.getContigIndex() == 1);
loc = iter.next();
assertTrue(loc.getStart() == 76);
assertTrue(loc.getStop() == 100);
assertTrue(loc.getContigIndex() == 1);
}
public GenomeLocSortedSet(GenomeLocParser parser, Collection<GenomeLoc> l) {
this(parser);
for (GenomeLoc e : l) add(e);
}
public GenomeLocSortedSet(GenomeLocParser parser, GenomeLoc e) {
this(parser);
add(e);
}