Cufflinks gtf file to htseq

Question

[AP]

Hi everyone,

I am trying to do differential gene expression analysis of fungal genome. For some reason I planned to switch my pipeline from trimmomatic/tophat/cufflink/... to trimmomatic/tophat/htseq/Dseq. However I am having problem using htseq because I don't have a GTF file. But then I realized that I have transcripts.gtf file from cufflinks output.

Can anyone tell me if its sensible to use the transcripts.gtf file from cufflink to htseq or Do I have any other option? Really confused and overwhlemed , please help me out.

Thank you, Ambika

Answer 1

[Kevin Blighe]

Hi Ambika,

From my perspective, it is a perfectly reasonable thing to do to use the transcripts.gtf file generated by Cufflinks for HTSeq. At the end of the day, HTSeq does not care which GTF file you provide - it just wants a well-formatted GTF file that matches the reference genome FASTA over which it will count read abundances.

What you have to ensure is that your logic for doing this is correct, i.e., by using a GTF file produced by Cufflinks, you are implying that you are searching for either novel transcripts in your fungal species of interest and/or novel splice-variants of existing transcripts.

In the past, with collaborators, I have performed experiments in this way using modified GTF files that contained hypothetical non-coding RNAs in cancer, and also employed something similar in a bacterial RNA-seq project.

Trust that this enlightens you!

Kevin

[AP]

Thank you Kevin for the clarification,

I was confused as I have different transcript.gtf file from each sample. Do you think I need to merge them before using or its ok to use separate ones for each sample?

[andrew.j.skelton73]

Use Cuffmerge to combine all your sample-wise GTF files into an experiment-wide GTF. I'd also recommend you tidy it up with TACO, especially if it's a denovo assembly.

[AP]

Thank you so much Andrew for your suggestion

[Kevin Blighe]

Yes, I agree with what Andrew says too