.nDiagnoses(500)

[pnrobinson]

Why are we doing 500 diagnoses?
Would it be simpler to limit this to nDiseases (usually 20)?

MaxoHpoDiseaseRank maxoHpoDiseaseRank = MaxoHpoDiseaseRank.Builder.builder()
                    .initialDiagnoses(allInitialDiagnoses)
                    .ascertainablePhenotypes(ascertainablePhenotypes)
                    .maxoToHpoTermIdMap(fullMaxoToHpoTermIdMap)
                    .maxoId(maxoId)
                    .sample(ppkt.getPpktSample())
                    .nDiagnoses(500)
                    .maxoLabel(biometadataService.maxoLabel(maxoId).get())
                    .build();

[pnrobinson]

I added this comment

    /* We sample HPO terms from the top 500 diagnoses.
    * This adds some HPO terms that correspond to the MAXO even though they
    * are not in the HPOAs of the top 20.
     */
    private final static int N_DIAGNOSES_FOR_SAMPLING_HPOS = 500;

and then

 MaxoHpoDiseaseRank maxoHpoDiseaseRank = MaxoHpoDiseaseRank.Builder.builder()
                    .initialDiagnoses(allInitialDiagnoses)
                    .ascertainablePhenotypes(ascertainablePhenotypes)
                    .maxoToHpoTermIdMap(fullMaxoToHpoTermIdMap)
                    .maxoId(maxoId)
                    .sample(ppkt.getPpktSample())
                    .nDiagnoses(N_DIAGNOSES_FOR_SAMPLING_HPOS)
                    .maxoLabel(biometadataService.maxoLabel(maxoId).get())
                    .build();

@mabeckwith we should discuss if we actually need this or if we should just sample from all of the terms related to the MAXO. This will be a bit easier to explain in the methods and might not change the results at all.