ClinVar should be a good start (which you can annotate with programs like ANNOVAR).
I think there is some room for improvement in terms of having access to the data to support those claims (and specific examples of evidence indicating the disease association in independent cohorts / families), but I think this should be a good way to start doing your own research to develop your opinion.
Also, it is nice that you have cases and controls in your own dataset (to control for differences in sample collection, sequencing design, and analysis).
That said, you may want to be careful amount formatting and compatibility of programs. For example, I am a cystic fibrosis carrier, and I had reports from some companies that falsely identified me as not being a carrier (even though you could verify the variant in the raw data):
https://github.com/cwarden45/DTC_Scripts/tree/master/Genos_Exome (please scroll to the bottom for alignment screenshot: also true for GET-Evidence report for Veritas WGS data)
In that example, that variant was an indel, and freebayes has a different format for indels (and 23andMe has a "D" and "I" notation in their raw format, rather than the sequence - even though the report within 23andMe was how I knew that I was a carrier).