/**
* @param protein : the blast protein
* @return the sequence coverage of the alignment for the match sequence
*/
public static float getMatchCoveragePercentFor(BlastResults protein) {
if (protein == null) {
return 0;
}
return ((float) (protein.getEndMatch() - protein.getStartMatch() + 1))
/ (float) protein.getSequence().length()
* 100;
}
private void writeBlastResults(
Writer writer, BlastResults blastResult, FastaSequence fastaSequence) throws IOException {
// candidate uniprot
writer.write(NEW_COLUMN);
writer.write(blastResult.getAccession());
// write identity
writer.write(NEW_COLUMN);
writer.write(Float.toString(blastResult.getIdentity()));
// write sequence coverages
writer.write(NEW_COLUMN);
writer.write(
Float.toString(getQueryCoveragePercentFor(blastResult, fastaSequence.getSequence())));
writer.write(NEW_COLUMN);
writer.write(Float.toString(getMatchCoveragePercentFor(blastResult)));
// write start/end
writer.write(NEW_COLUMN);
writer.write(Integer.toString(blastResult.getStartQuery()));
writer.write("-");
writer.write(Integer.toString(blastResult.getEndQuery()));
writer.write(NEW_COLUMN);
writer.write(Integer.toString(blastResult.getStartMatch()));
writer.write("-");
writer.write(Integer.toString(blastResult.getEndMatch()));
// write query sequence
writer.write(NEW_COLUMN);
writer.write(fastaSequence.getSequence());
// write alignment
writer.write(NEW_COLUMN);
writer.write(blastResult.getAlignment());
// write matching sequence
writer.write(NEW_COLUMN);
writer.write(blastResult.getSequence());
writer.write(NEW_LINE);
writer.flush();
}
/**
* Checks that the sequence of the BlastProtein is not in conflict with the feature ranges that
* the protein can have in Intact
*
* @param range : the range of the feature to check
* @param protein : the protein hit which could replace the old protein in Intact
* @return true if there is no conflict between the sequence of this BlastProtein and the range
*/
private boolean checkRangeValidWithNewSequence(
Range range, BlastResults protein, BlastReport report) {
// The difference between the previous start position and the new one in the new sequence
int diffStart = protein.getStartQuery() - protein.getStartMatch();
// The difference between the previous end position and the new one in the new sequence
int diffEnd = protein.getEndQuery() - protein.getEndMatch();
// Shift the ranges in consequence
int startFrom = range.getFromIntervalStart() - diffStart;
int startTo = range.getToIntervalStart() - diffStart;
int endFrom = range.getFromIntervalEnd() - diffStart;
int endTo = range.getToIntervalEnd() - diffStart;
// No ranges should be before the new start positions
if (startFrom < protein.getStartMatch()
|| range.getFromIntervalStart() < protein.getStartQuery()) {
report.addWarning(
"The feature range is "
+ range.getFromIntervalStart()
+ "-"
+ range.getToIntervalStart()
+ " and the alignment with the Swissprot sequence starts after "
+ range.getFromIntervalStart()
+ ". We can't change the previous sequence with the sequence of the Swissprot entry because it will be incoherent with the current feature(s) of the protein.");
return false;
}
// No ranges should be before the new start positions
else if (startTo > protein.getEndMatch()
|| range.getToIntervalStart() > protein.getEndQuery()) {
report.addWarning(
"The feature range is "
+ range.getFromIntervalStart()
+ "-"
+ range.getToIntervalStart()
+ " and the alignment with the Swissprot sequence finishes before "
+ range.getToIntervalStart()
+ ". We can't change the previous sequence with the sequence of the Swissprot entry because it will be incoherent with the current feature(s) of the protein.");
return false;
}
// No ranges should be after the new end positions
else if (endFrom < protein.getStartMatch()
|| range.getFromIntervalEnd() < protein.getStartQuery()) {
report.addWarning(
"The feature range is "
+ range.getFromIntervalEnd()
+ "-"
+ range.getToIntervalEnd()
+ " and the alignment with the Swissprot sequence starts after "
+ range.getFromIntervalStart()
+ ". We can't change the previous sequence with the sequence of the Swissprot entry because it will be incoherent with the current feature(s) of the protein.");
return false;
}
// No ranges should be after the new end positions
else if (endTo > protein.getEndMatch() || range.getToIntervalEnd() > protein.getEndQuery()) {
report.addWarning(
"The feature range is "
+ range.getFromIntervalEnd()
+ "-"
+ range.getToIntervalEnd()
+ " and the alignment with the Swissprot sequence finishes before "
+ range.getToIntervalStart()
+ ". We can't change the previous sequence with the sequence of the Swissprot entry because it will be incoherent with the current feature(s) of the protein.");
return false;
} else {
String rangeSequence = range.getFullSequence();
if (rangeSequence == null) {
rangeSequence = range.getFullSequence();
}
// Check that the amino acids involved in the feature ranges are identical
if (startFrom > 0 && endTo > 0) {
String rangeNewSequence = protein.getSequence().substring(startFrom - 1, endTo);
if (rangeSequence == null) {
report.addWarning(
"The feature "
+ range.getFeature().getAc()
+ " doesn't contain any sequence or full sequence but has range positions. This entry needs to ce checked by a curator");
return false;
}
if (!rangeSequence.equals(rangeNewSequence)) {
report.addWarning(
"The sequence of the Swissprot entry from "
+ range.getFromIntervalStart()
+ " to "
+ range.getToIntervalStart()
+ " is different from the previous feature sequence, we can't replace the previous sequence with the sequence of the Swissprot entry.");
return false;
}
}
}
return true;
}