Recommended upstream alignment

[holtjma]

I was tinkering with the software, mainly to see ease-of-use and ran into an issue with regards to upstream processing. I have two alignment processes that are almost identical (sentieon align, dedup, BQSR) except one performs postalt correction and the other does not. When I run on three NA12878 replicates I get the following results: No-postalt:

Sample  Genotype
SL362490    *3/*4+*68
SL362491    *3/*4+*68
SL362492    *3/*4+*68

Postalt+:

Sample  Genotype
SL362490    None
SL362491    None
SL362492    None

The difference is obvious, which leads to the question of what's recommended for upstream alignment? Will performing postalt processing always lead to None or is there some workaround that will fix that?

[xiao-chen-xc]

Hi @holtjma

I tried postalt and it changes the majority of reads in CYP2D6 from a good MapQ to a MapQ of 0, due to equally-good alignments to some ALT contigs. This causes problem to Cyrius as it uses those good reads to call variants. I think the most straightforward solution would be to use the bams before postalt to run Cyrius. If that's not feasible, we can think of some workaround to work with postalt-processed bams.

Thanks, Xiao

[holtjma]

Thanks for the response, I suspected it was something along those lines. Post-alt processing seems to be a mixed bag of good and bad, so I'm not too surprised that it would cause an issue in this particular context.

As for the workaround, yea that probably isn't feasible for a long-term solution. Post-alt is usually done on read-sorted BAMs (and then sorted afterwards), and I think Cyrius requires position sorted based on the speed. It's obviously doable, just redundant in some aspects. I'll continue testing with some BAMs processed without post-alt for now.

On the plus side, the tool seemed to work right out of the box on non-postalt and was quite fast!

[xiao-chen-xc]

Glad to know it's working well for you. Feel free to reach out to me if you run into any other problem.