/**
* Is this single change a LOF?
*
* <p>Criteria: 1) Core splice sites acceptors or donors (only CORE ones) 2) Stop gained (if this
* happens at the last part of the protein, we assume it has no effect) 3) Frame shifts
*
* @param changeEffect
* @return
*/
protected boolean isNmd(ChangeEffect changeEffect) {
Transcript tr = changeEffect.getTranscript();
if (tr == null)
throw new RuntimeException("Transcript not found for change:\n\t" + changeEffect);
// Only one exon? Nothing to do (there is no exon-exon junction)
if (tr.numChilds() <= 1) return false;
// Find last valid NMD position
int lastNmdPos = lastNmdPos(tr);
if (lastNmdPos < 0) return false; // No valid 'lastNmdPos'? => There is no NMD event.
// Does this change affect the region 'before' this last NMD position? => It is assumed to be
// NMD
Variant seqChange = changeEffect.getSeqChange();
boolean nmd;
if (tr.isStrandPlus()) nmd = seqChange.getStart() <= lastNmdPos;
else nmd = lastNmdPos <= seqChange.getEnd();
// Update sets and counters
if (nmd) {
transcriptsNmd.add(
changeEffect
.getTranscript()); // Unique transcripts affected (WARNING: null will be added)
genesNmd.add(changeEffect.getGene()); // Unique genes affected (WARNING: null will be added)
nmdCount++;
}
return nmd;
}
/**
* Is this deletion a LOF?
*
* <p>Criteria: 1) First (coding) exon deleted 2) More than 50% of coding sequence deleted
*
* @param changeEffect
* @return
*/
protected boolean isLofDeletion(ChangeEffect changeEffect) {
Transcript tr = changeEffect.getTranscript();
if (tr == null)
throw new RuntimeException("Transcript not found for change:\n\t" + changeEffect);
// ---
// Criteria:
// 1) First (coding) exon deleted
// ---
if (changeEffect.getEffectType() == EffectType.EXON_DELETED) {
Variant seqChange = changeEffect.getSeqChange();
if (seqChange == null)
throw new RuntimeException("Cannot retrieve 'seqChange' from EXON_DELETED effect!");
if (seqChange.includes(tr.getFirstCodingExon())) return true;
}
// ---
// Criteria:
// 2) More than 50% of coding sequence deleted
// ---
// Find coding part of the transcript (i.e. no UTRs)
Variant seqChange = changeEffect.getSeqChange();
int cdsStart = tr.isStrandPlus() ? tr.getCdsStart() : tr.getCdsEnd();
int cdsEnd = tr.isStrandPlus() ? tr.getCdsEnd() : tr.getCdsStart();
Marker coding = new Marker(seqChange.getChromosome(), cdsStart, cdsEnd, 1, "");
// Create an interval intersecting the CDS and the deletion
int start = Math.max(cdsStart, seqChange.getStart());
int end = Math.min(cdsEnd, seqChange.getEnd());
if (start >= end) return false; // No intersections with coding part of the exon? => not LOF
Marker codingDeleted = new Marker(seqChange.getChromosome(), start, end, 1, "");
// Count:
// - number of coding bases deleted
// - number of coding bases
int codingBasesDeleted = 0, codingBases = 0;
for (Exon exon : tr) {
codingBasesDeleted += codingDeleted.intersectSize(exon);
codingBases += coding.intersectSize(exon);
}
// More than a threshold? => It is a LOF
double percDeleted = codingBasesDeleted / ((double) codingBases);
return (percDeleted > deleteProteinCodingBases);
}