cufflinks not able to assemble transcripts in high read coverage area

[apratap]

Hey Guys

I am using cufflinks for building transcriptome. I am seeing some regions where in-silico model predictions is re-affirmed by RNA-Seq data but however cufflinks or scripture is not able to assemble a transcript.

Is there anything that we can do to improve the sensitivity of cufflinks.

Here is a snapshot of on such example.



PS: this message was also cross posted on Biostar but unfortunately I didn't get any response there.

Thanks!
-Abhi

[Wallysb01]

You can increase the -L option to filter out a higher isoform fraction. This should help with not including unspliced introns. Also, increasing -j will help the same thing.

Though its not as related to your specific question, I've increased the --min-frags-per-transfrag option to reduce the number of transcripts called.

I've found this RABT assembly to be pretty messy no matter what parameters I fiddled with. If you're really concerned with creating the best annotations you can, you should probably use ABySS and trans-ABySS. Its a lot of work, but I'm starting to get some very good results.

[apratap]

@Wallysb01 : Thanks for your reply.

I am not sure how -j could help as the number of reads that seem to mapping to intronic regions for this particular gene is very less compared to spliced reads.

Also wats a RABT assembly ?

I am interested in knowing any specific reason why you find ABySS and trans-ABySS to do a better job ?

Thanks!
-Abhi