Pre-filtering of annotated isoforms based on splice junctions

Question

[I0110]

Is there a method to exclude transcript isoforms without any splicing junction support from RNA-seq data? I would like to perform some pre-filtering of annotated isoforms before using Kallisto or RSEM. Thanks!

[dthorbur]

You could parse the reference genome GTF/GFF file and count the number of exons per transcript (using the type column). Then any transcript with only 1 annotated exon could be removed from the reference transcriptome.

However, I would be wary of removing references before mapping. Reads that initially mapped to single-exon transcripts may map elsewhere inflating read counts. For example, this could be true for genes evolving through duplication.