Hi All,
I know there is a discussion regarding difference between gene and transcript expression at BioStar. However, I am still confused and I would like to open this discussion again. I have a RNA-seq data (analyzed by kallisto/Sleuth) which shows significant changes in gene expression at transcript level but not at genes level. In other word, e.g. gene x is significantly and highly expressed at transcript-RA but not transcript-RB or transcript-RC. When I looked at the gene expression of gene x at gene level the expression is low and not significant. So, my question is how the gene expression considered? (I thought the gene expression is a sum of transcripts expression of gene x, but based on my observation it is not!). How gene expression computed at gene level?
Thanks Morteza
It depends on how you did your transcript- to gene-level summarization, which you didn't tell us how you performed. I am not familiar with kallisto and Sleuth, do they provide some method or function to do such summarization? I am aware of tximport, in which case I believe gene-level should be (approximately) the sum of the estimated counts for all transcripts from that gene.
I think the biggest question is how to decide the same exon belongs to different transcript. Then you will got it. Hope helpful.
kallisto uses an EM algorithm to distribute reads along different transcripts sharing exons.