masked or unmasked genome reference for RNAseq

[skly]

Hi guys. I aligned RNAseq data to unmasked genome and hard masked genome, respectively. The mapping rate of the the former was about 92% and the latter rate was about 75%. It seemed that about 20% RNAseq data aligned to the repeating sequences of genome. So, which genome reference was better for RNAseq data, the unmasked genome or the hard masked?
Thanks a lot!

[dpryan]

Use the unmasked (or soft masked) genome. Actually, use that for everything that doesn't explicitly state that it wants a hard masked genome. There are MANY genes that overlap repeat regions, at least partially and you'll be missing alignments to them if you hard mask a genome. Similarly, there's often expression of some repeats (this is mostly just noise), and by using a hard masked genome you'll increase your false-positive alignment rate of sequence originated from such repeats.

[skly]

Thanks dpryan~
I am clear.

Use the unmasked (or soft masked) genome. Actually, use that for everything that doesn't explicitly state that it wants a hard masked genome. There are MANY genes that overlap repeat regions, at least partially and you'll be missing alignments to them if you hard mask a genome. Similarly, there's often expression of some repeats (this is mostly just noise), and by using a hard masked genome you'll increase your false-positive alignment rate of sequence originated from such repeats.