I am looking for the best approach to predict novel and alternative splicing events from RNA-seq data. I tried to use Cufflink and Trans-AByss so far, however they both have limitations that makes using them difficult. Cufflink requires huge amount of memory and takes weeks to run (And I had only 20 human transcriptome!). While Trans-AByss does not predict the expression level (ex. FPKM) for transcripts, and has high false positive rate! I have also heard about this method called RSEM (going to try it), but I think it only quantify known transcripts from RNAseq data.
So I think I am wondering what is the best approach for identification and quantification of both novel and alternative splicing events in your experience?
Thanks very much :-)