[miso-users] Question about assigned counts

[Wang, Ziheng]

Dear MISO users,

I have noticed that the numbers of the assigned counts do not conform to the probability distribution described by the psi values for the isoforms. I am aware that this should happen but I am just a bit confused about what exactly the difference entails: is it just that the psi values account for isoform length, and in pair-end cases, insert length as well?

Sometimes, the values diverge quite a bit: for example, on a run on a gene with 17 isoforms, the psi values for two isoforms were 0.40 and 0.43, whereas their assigned reads were 2302 and 490 respectively. The lengths were indeed different but not by that much: 520 amino acids versus 357 amino acids. (ENSG00000187098, ENST00000352241 and ENST00000531774, on ERR030857 ) How can the shorter isoform be favored despite having 25% of the assigned counts? 

Thank you,

Ziheng Wang