any way to use RNA seq mapping to choose between competing gene models?

Question

[sullis02]

I have run gene prediction with several different parameter sets, and the different annotation outputs that result include different gene length and gene models for the same genomic regions. I also have a good .bam file from STAR mapping of RNAseq reads to the genome . The RNAseq mapping was independent of gene prediction (i.e. mapping was not annotation guided).

Short of visually examining the RNA seq mapping to every set of overlapping gene models, to find the most 'correct' one, is there software that can leverage the bam file to do it algorithmically? Take a set of overlapping gene model coordinates and choose the one best supported by RNAseq evidence?

I realize this would only 'work' for expressed genes, but it's still a start.

[GenoMax]

Have you checked sashimi plots in IGV (LINK)? Also in MISO (LINK).