As an alternative to step 2, you could also score the filtered vcf file (this is key to get rid of most of the initially millions of variants) with a variant prioritization framework like CADD (cadd.gs.washington.edu) to immediately highlight 'interesting' variants. Highlighting certain genes can definetly help, though it is not always conclusive

(I do not want to demotivate but please be cautious. Even the best analysts have 'only' around 50% case-solve rate. If it is an adult-onset-disease, chances can be lower as the disease mechanism in that case may not be 'consequential' enough to be naturally selected against)