static List<Feature> computeGeneLevelExpression(
Map<String, FeatureCount> featureCounts,
Map<String, Feature> geneInfos,
double numberOfReads) {
List<Feature> geneFeatures = new LinkedList<Feature>();
HashMap<String, List<FeatureCount>> geneFeaturesMap = new HashMap<>();
// Organize exons by gene
Collection<FeatureCount> exons = featureCounts.values();
for (FeatureCount fc : exons) {
String geneId = fc.getFeature().getAttribute("gene_id");
if (geneFeaturesMap.containsKey(geneId)) {
geneFeaturesMap.get(geneId).add(fc);
} else {
LinkedList<FeatureCount> fcs = new LinkedList<>();
fcs.add(fc);
geneFeaturesMap.put(geneId, fcs);
}
}
// compute total counts and total length
for (Map.Entry<String, List<FeatureCount>> entry : geneFeaturesMap.entrySet()) {
List<FeatureCount> fcsForGene = entry.getValue();
Collections.sort(fcsForGene);
Feature geneInfo = geneInfos.get(entry.getKey());
if (geneInfo == null) {
throw new IllegalStateException(entry.getKey() + " has no entry in geneInfos");
}
if (!geneInfo.type().equals("gene")) {
throw new IllegalStateException(geneInfo + " is not a gene");
}
// get total length and counts
double count = 0.0;
int length = 0;
List<String> exonIds = new LinkedList<>();
for (FeatureCount fc : fcsForGene) {
count += fc.getCount();
length += fc.getFeature().location().length();
exonIds.add(fc.getId());
}
// comput rpkm
double rpkm = count / length / numberOfReads * Math.pow(10, 9);
HashMap<String, String> info = new HashMap<>();
info.put("reads", Double.toString(count));
info.put("RPKM", Double.toString(rpkm));
info.put("length", Integer.toString(length));
info.put("exons", StringUtils.join(exonIds, ","));
Feature newGeneInfo = GTFFeatureBuilder.addAttributesToFeature(geneInfo, info);
geneFeatures.add(newGeneInfo);
}
return geneFeatures;
}
@Override
public List<Feature> getCounts() {
List<Feature> newFeatures = new LinkedList<>();
for (FeatureCount fc : this.featureCounts.values()) {
Feature f = fc.getFeature();
Feature gene = geneInfo.get(f.getAttribute("gene_id"));
int tss =
gene.location().bioStrand() == '+'
? gene.location().bioStart()
: gene.location().bioEnd();
Map<String, String> newAtts = new HashMap<>();
newAtts.put("id", fc.getId());
newAtts.put("reads", Double.toString(fc.getCount()));
newAtts.put("RPKM", Double.toString(fc.getRPKM(this.getTotalCount())));
newAtts.put("tss", Integer.toString(tss));
Feature newF = GTFFeatureBuilder.addAttributesToFeature(f, newAtts);
newFeatures.add(newF);
}
return newFeatures;
}