Transcript to gene mapping of overlapping transcripts from cuffcompare

Question

[sunil.mangalam]

Hi Forum, We did RABT assembly of mouse transcriptome using cufflinks and stringtie. Cuffcompare produced many genes of the class code O. While mapping these transcripts to gene level, cuffcompare grouped transcripts of class code o to the parent gene. But many o transcripts look very different from the gene it was grouped with and it is justified to call them as novel genes. Has any one looked at this problem and deviced criteria to classify o class transcripts? I think cuffcompare's criteria does not work well. In fact, I saw all transcripts with classcode X (antisense transcripts) grouped with the known gene, but this does not make sense, as they are transcribed from opposite strands. Thanks! Sunil