You can use algorithms to prioritize the variants in the VCF file based on how damaging they are predicted to be, for example VAAST http://www.yandell-lab.org/s

You can use algorithms to prioritize the variants in the VCF file based on how damaging they are predicted to be, for example VAAST http://www.yandell-lab.org/software/VAAST/VAAST_Quick-Start-Guide.pdf http://www.yandell-lab.org/software/VAAST/VAAST_Quick-Start-... or Exomiser. There are ways to bias the search using phenotypes that you know of. The issue with using existing databases is that you will never find a novel variant that may be causative. There are companies that provide this service (e.g. Fabric Genomics) and they may be willing to help.