About gap extension decay

[SHuang-Broad]

Hi Marcus,

I'm wondering what the intuitions are behind the gap extension decay parameter?

In particular, I've noticed the use of that parameter is modeled as

 min(0, self.gap_extension_penalty + ins_run * self.gap_extension_decay)

and similarly for deletions.

My guess is that this is to further discourage large gaps? Is there a specific use scenario you had in mind for this model?

Thanks! Steve

[mbreese]

The idea would be that for RNAseq data, you might want allow for the potential for skipping introns (or gaps). So that if you once you hit a deletion of a certain size, you might want to reduce the penalty.

So, for example, you might allow a deletion of 10bp, but after that, you’d have no penalty.

I’m not sure how practical it ends up being in practice though.

— Marcus

On May 5, 2021, at 2:41 PM, Steve Huang @.***> wrote:

 Hi Marcus,

I'm wondering what the intuitions are behind the gap extension decay parameter?

In particular, I've noticed the use of that parameter is modeled as

min(0, self.gap_extension_penalty + ins_run * self.gap_extension_decay) and similarly for deletions.

My guess is that this is to further discourage large gaps? Is there a specific use scenario you had in mind for this model?

Thanks! Steve

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[SHuang-Broad]

Ah, I see. I assumed it was for DNA. Thanks for the quick answer! Steve