how do you deal with "final" NGS resequencing results? or in other words, how exactly do you solve the always uncomfortable "I've got thousands of variants, now what?" question?
one is sometimes surprised how almost all the bfx efforts are focused on generating results rather than interpreting them. in this NGS era, sure aligners and variant callers must be created and optimized, but if we don't have a fast and reliable way of dealing with their results these will be not as useful as we'd like. it makes sense for me that first you need these mappers and callers, but now that labs have already generated plenty of NGS data results interpretation becomes critical in order to give NGS techniques the proper credit they deserve.
there are not many ways one can deal with NGS variants. basically you try to annotate your variants as best as possible, so you can then perform a cascade of filters that would allow you to clean your relevant findings out from the NGS noise. but if you care a lot about what you are filtering and what you are looking for, even the most obvious filter of novel variants could be argued. filters are definitely helpful, but we shouldn't stop there if our aim is to discover all the answers, and not only a few of them.
a little less conversation, a little more action please.
we are currently using a SOLiD sequencer and its own software to map reads and to call variants, and we are enriching them using ANNOVAR (the most convenient annotation software we've found to date). then we are handling these annotated results to the biologists and clinicians for them to apply their knowledge (because we think that variant filtering should be expert guided), and eventually something is found and a paper is written ;)
we wouldn't like to stop here, as manual (or semi-automated) filtering could be error prone or simply not properly applied, so we would like to know how others are dealing with this situation, specially in terms of easing the variant selection process on expert non-bfx hands and also of generating graphical results for visual analysis and paper dressing. we know there are a few alternatives out there for variant interpretation (the only ones we've played a little bit with were Partek and the recently published SVA), so it would be great to get here some feedback here about the ones you've used, just tested or even discarded.