Finding Novel Splicing Events/Transcripts Using Tophat And Cufflinks

Question

[Sahel]

Hi There,

I would like to identify novel splicing events occurring in 4 human paired-end RNA-seq samples. From literature I figured tophat and cufflink can do such thing. So I used Tophat to assemble and Cufflink to find all transcripts. Next I used cuffcompare to identify novel transcripts from known ones (using gene.gtf downloaded from UCSC table browser). And then I got the ones with class_code = "j", which according to manual should be novel.

So now I am left with a list like this:

chr1    Cufflinks       exon    885636  886043  .       +       .       gene_id "XLOC_000025"; transcript_id "TCONS_00000033"; exon_number "1"; gene_name "ENST00000379410"; oId "CUFF.20.1"; nearest_ref "ENST00000379410"; class_code "j"; tss_id "TSS33";
chr1    Cufflinks       exon    886536  887714  .       +       .       gene_id "XLOC_000025"; transcript_id "TCONS_00000033"; exon_number "2"; gene_name "ENST00000379410"; oId "CUFF.20.1"; nearest_ref "ENST00000379410"; class_code "j"; tss_id "TSS33";
chr1    Cufflinks       exon    887947  888496  .       +       .       gene_id "XLOC_000025"; transcript_id "TCONS_00000033"; exon_number "3"; gene_name "ENST00000379410"; oId "CUFF.20.1"; nearest_ref "ENST00000379410"; class_code "j"; tss_id "TSS33";
chr1    Cufflinks       exon    888580  888747  .       +       .       gene_id "XLOC_000025"; transcript_id "TCONS_00000033"; exon_number "4"; gene_name "ENST00000379410"; oId "CUFF.20.1"; nearest_ref "ENST00000379410"; class_code "j"; tss_id "TSS33";
chr1    Cufflinks       exon    889163  889251  .       +       .       gene_id "XLOC_000025"; transcript_id "TCONS_00000033"; exon_number "5"; gene_name "ENST00000379410"; oId "CUF

But I do not know how to interpret them? How I can get the sequence of the novel transcripts? Is there a way to figure what happened that result in a novel junction? like if it is an insertion, deletion and if cause frameshift?

This is my first time doing such analysis, any help would be greatly appreciated, :-)

Sahel

Answer 1

[Mikael Huss]

You can get the sequences of the novel transcripts (or the exons, to be more precise) with a tool like BEDTools ("bedtools getfasta"). It can accept BED, GFF and other formats and might work straight away on your example, which looks like GFF.

For the other questions, I don't have as straightforward a reply. I would import the regions to some genome browser like UCSC Genome Browser or IGV, compare them to existing annotation and take it from there.