`diseaseMatches` is not populated in parquet output but `HUMAN_PHENO_EVIDENCE` populated in genes.tsv output

[yaseminbridges]

When regenerating genes.tsv from Exomiser parquet results, HUMAN_PHENO_EVIDENCE is empty for some genes despite a non-zero diseasePhenotypeScore. This occurs because the diseaseMatches field is not populated in the parquet for these genes, even though the corresponding evidence is present in the original Exomiser TSV output.

Expected behaviour

diseaseMatches should be populated wherever the original genes.tsv contains human phenotype evidence.

Actual behaviour

diseaseMatches is empty in the parquet for affected genes, making it impossible to reproduce HUMAN_PHENO_EVIDENCE from the parquet alone.

Example

I have attached the sample used to run with Exomiser v15.0.0, along with the corresponding VCF. I observed this with genes PLG and PAH.

Asgharzade-2018-GIPC3-Ahv-14_23.json

Asgharzade-2018-GIPC3-Ahv-14_23.vcf.gz