private void mergeProfiles(int p100Length, int gcpLength) {
log.info("Merging profiles");
ArrayList<AssayType> dummyAssay = new ArrayList<>();
dummyAssay.add(AssayType.GCP);
dummyAssay.add(AssayType.P100);
List<Profile> profiles = profileRepository.findByAssayTypeInOrderByConcatDesc(dummyAssay);
String prevConcat = null;
String curConcat;
List<Profile> bunchOfProfiles = null;
for (Profile profile : profiles) {
curConcat =
profile.getReplicateAnnotation().getCellId()
+ profile.getReplicateAnnotation().getPertiname();
if (prevConcat == null) {
prevConcat = curConcat;
bunchOfProfiles = new ArrayList<>();
}
if (!curConcat.equals(prevConcat)) {
prevConcat = curConcat;
MergedProfile mergedProfile =
UtilsTransform.mergeProfiles(bunchOfProfiles, p100Length, gcpLength);
mergedProfileRepository.save(mergedProfile);
bunchOfProfiles = new ArrayList<>();
} else {
bunchOfProfiles.add(profile);
}
}
}
private void normalize() {
List<AssayType> assayTypes = Arrays.asList(AssayType.GCP, AssayType.P100);
// do loop for each assay e.g. P100, GCP
for (AssayType assayType : assayTypes) {
log.info("Normalize matrix of peak values for assay: {}", assayType);
List<PeakArea> allPeakAreas = peakAreaRepository.findByGctFileAssayType(assayType);
// inefficient, try with database count
int numberOfPeptides = peptideAnnotationRepository.findByAssayType(assayType).size();
int numberOfReplicates = replicateAnnotationRepository.findByAssayType(assayType).size();
ArrayList<Integer> mapPeptideIdToRowId = new ArrayList<>();
ArrayList<Integer> mapReplicateIdToColumnId = new ArrayList<>();
// double[][] matrix = new double[replicates][peptides];
// init matrix with sizes
List<List<Double>> peaksAsMatrix = new ArrayList<>(numberOfReplicates);
for (int i = 0; i < numberOfReplicates; i++) {
peaksAsMatrix.add(new ArrayList<>(numberOfPeptides));
}
for (PeakArea peakArea : allPeakAreas) {
int peptideId = Math.toIntExact(peakArea.getPeptideAnnotation().getId());
int replicateId = Math.toIntExact(peakArea.getReplicateAnnotation().getId());
Double rawValue = peakArea.getValue();
int mappedRowId;
int mappedColumnId;
if (!mapPeptideIdToRowId.contains(peptideId)) {
mapPeptideIdToRowId.add(peptideId);
}
mappedRowId = mapPeptideIdToRowId.indexOf(peptideId);
if (!mapReplicateIdToColumnId.contains(replicateId)) {
mapReplicateIdToColumnId.add(replicateId);
}
mappedColumnId = mapReplicateIdToColumnId.indexOf(replicateId);
peaksAsMatrix.get(mappedColumnId).set(mappedRowId, rawValue);
}
List<List<Double>> outputMatrix = Normalizer.quantileAndZScoreNormalize(peaksAsMatrix);
// write normalized values back to DB
for (int i = 0; i < numberOfReplicates; i++) {
for (int j = 0; j < numberOfPeptides; j++) {
Double normalizedValue = outputMatrix.get(i).get(j);
Long databaseReplicateId = mapReplicateIdToColumnId.get(i).longValue();
Long databasePeptideId = mapPeptideIdToRowId.get(j).longValue();
List<PeakArea> peakAreas =
peakAreaRepository
.findByGctFileAssayTypeAndReplicateAnnotationIdAndPeptideAnnotationId(
assayType, databaseReplicateId, databasePeptideId);
assert peakAreas.size() == 1;
PeakArea peakArea = peakAreas.get(0);
peakArea.setNormalizedValue(normalizedValue);
// Fill in normalized value field in DB
peakAreaRepository.save(peakArea);
}
}
log.info(
"Normalized assay: {} peptides: {}, replicates: {}.",
assayType,
numberOfPeptides,
numberOfReplicates);
}
}
