Identifying intron sequences in de novo assembled transcripts?

[BobFreemanMA]

Greetings!

We're using Trinity to do de novo assembly of frog RNA-seq data. From the resulting assemblies, I'd like to be able to identify any transcripts that contain intronic sequences (from partially-spliced transcripts) so that we can filter these out of our data set. Does anyone know of a program that might do this? It would not be a problem for me to put together sets of sequences that represent introns, coding exons, and non-coding exons, in case I need to train a program, e.g. a gene predictor. I fear that predictors will not work well or at all on sequences that are the size of transcripts.

Thanks,
Bob

[pengchy]

If your RNAseq data is produced with poly-A enrichment, the sequence should be all exons, regardless coding or non-coding. How to define the intron region for this conditioin?

[BobFreemanMA]

In theory, this is true. However, RNA-seq is so sensitive that it is likely one is also capturing pre-spliced and partially spliced mRNAs, although in very low %s. These species of molecules are the ones I'm trying to identify and remove.