mapreads and IUB mapping

[OneManArmy]

Has anyone tried using mapreads with a IUB coded reference? I am having problems here because there are less mappings with the IUB coded reference than a reference without IUB codes.

[westerman]

Short answer: no.

Longer answer: I didn't think IUB codes in the reference -- and thus mapreads -- were allowed. The corona lite manual mentions that the sequence part of the reference "should contain only A, C, G, N, T, and newline" characters. Uppercase only. The program 'reference_validation.pl' will create a valid sequence.

But I am not running the latest and greatest Corona Lite version yet.

[nilshomer]

Has anyone tried using mapreads with a IUB coded reference? I am having problems here because there are less mappings with the IUB coded reference than a reference without IUB codes.

What application are you trying to use IUB codes with? For example, in Human data there should not be too big of a hit on coverage if you don't use IUB codes (informed say from dbsnp) as long as your sensitivity is high (most aligners allow you to increase the sensitivity at the cost of some speed).

I think the lack of support of IUB codes (from my own experience as an alignment algorithm designer) is that there is a significant speed decrease to support them, whereas sensitivity can overcome this (I don't know if the speed decrease is the same).

[OneManArmy]

This is a very specific application with a very small reference - AB's documentation shows there is a 30% decrease in performance with IUB enabled, and this is acceptable for my use.

I tried using the newest mapreads on solidsoftwaretools.com, and also with the mapreads that comes with SATS 3.0 (they are the same version 1.1.9). There is some mention of IUB codes with the matching random pipeline, so I thought it would work but obviously it is buggy.