Comparing Libraries With Different Read Lengths

Question

[CF]

Greetings,

I am interested in testing for differential expression between a number of conditions. I usually use both DESeq2 and Cuffdiff to achieve this. Some of our samples come from different batches, for which I already attempt to correct. The libraries are from paired RNASeq data with read lengths of 100bp. However, I was recently given data sets with read lengths of 150bp (newer protocol). Some are replicates of existing data (we have both 100bp and 150bp reads), but others belong to a newly obtained sample group. Changing the read length will surely change their mappability.

Can the 100bp data be compared to the 150bp data? Can differences in read length be corrected for, either by trimming reads or during DE testing?

Thanks!

[AntiSocialBehaviourOrder]

Maybe this paper will be of assistance

Answer 1

[Devon Ryan]

If you're already trying to correct for a batch effect then I suppose you can just include read length as part of the batch. Having said that, please do be careful with batch correction and try to make sure that it's actually worked.

There's no reason to use cuffdiff.