As far as I am aware, none of the standard transcript assembly tools (cufflinks/stringtie etc) have the ability to annotate the open reading frame within a denovo assembled transcript, they simply say which part of a the genome is transcribed and which is not.

We have attempted to do this in the past in two different ways. Neither approach is 100% perfect. I can also think of a third approach that might work, but we havn't tried.

1. Obtain the sequence of each of the newly assembled transcripts, probably as a Fasta file, and either write a script or use one of the existing tools, to find the first or longest ORF within the sequence. The UTRs are then the sequences before or after this location. E.g. https://www.ncbi.nlm.nih.gov/orffinder/ or https://stackoverflow.com/questions/31757876/python-find-longest-orf-in-dna-sequence
2. If you have an already annotated reference transcriptome, and you are just looking for additional transcripts, for each novel transcript you can look for a reference transcript with an identical intron chain within the coding sequence, and whose ATG and stop codon is the within the novel transcript and transfer over the coordinates
3. If there are protein annotations for this, or a related speices, you can use tblastn to align your novel transcript sequences against the protein database, and identify the open reading frames that way.