Dear biofellas, I have a phylosophical question here. While on one hand I am aware that the title question can be either supported or not by each one of us, I am interested in gathering your opinions to make a consensus of them and make a choice for my project :)
I have five RNASeq libraries from five different species, the insert size distributions seem "normal" (both in -commonsense- normal and -gaussian- normal), but I want to improve mapping accuracy a bit. I am mapping on the transcriptome of one of the five species, so I am not using splice-aware methods. The fact that the species are different obviously alters the insert size distribution a bit itself.
Do you think it makes sense to restrict the insert size range between the values values that my bell-shaped plots show? Or should I instead make use of everything I find, because of the difference in terms of species between reads and reference?
Disclaimer: I am 1% more confident in not considering the insert size.