Input from one of our physician/geneticists, Dr. Leslie Manace. Unfortunately, I do not have a specific url/location/resource to point this comment to. I believe the WG will be able to consider this generally and apply as appropriate. Fortunately, Kevin Power was able to provide inital feedback, which I have included below.

Dr Manace: Genetic Interpretations - I would like to see more sub-categorization for sequence variants - for these (vs. deletion/duplication variants), supporting information is usually critical for their interpretation as pathogenic vs. unknown significance (there are very few single site mutations that are recurrent and so well described that they speak for themselves). There is an ACMG variant interpretation algorithm (attached, see p.10 for rules) that I'd want to see incorporated.

Kevin Power:

"I would like to see more sub-categorization for sequence variants - for these (vs. deletion/duplication variants)"

We originally had more specific ones, but instead we landed on having an attribute (as one of those "Components" that describe the type of variation (like deletion, duplication, etc …). No one was able to articulate how different the supporting information needed to be for those different types to warrant breaking up any further than we have today. But, that doesn't mean we shouldn't.

"supporting information is usually critical for their interpretation as pathogenic vs. unknown significance (there are very few single site mutations that are recurrent and so well described that they speak for themselves)"

I think we might revisit how we structure our genetic impacts (http://hl7.org/fhir/uv/genomics-reporting/general.html#impacts ) to make sure we support ACMG recommendations. I think it might be close today, but I haven't compared the most recently iterations of our guide to ACMG, as towards the end we were a little more focused on Somatic.

"There is an ACMG variant interpretation algorithm (attached, see p.10 for rules) that I'd want to see incorporated"

We would not support "executing" the algorithm, but instead would want to make sure we have a structure in place to hold the interps coming out of an algorithm like that - so that those interps can be shared in a consistently structured fashion.