Cufflinks input- reference information

Question

[Lindsay44]

I am looking for novel genes in two different experimental groups ( differential expression analysis).

When do I include the following information in my command?

Cufflinks- reference annotation (GTF)

Cuffmerge- reference annotation (GTF) and reference sequence (Fasta)

Cuffdiff- reference GTF

Thanks for the help.

Answer 1

[Satyajeet Khare]

Reference GTF in tophat, cufflinks, cuffmerge and merged.gtf in cuffdiff will differentiate novel genes from known and therefore need to be included if you want known as well as novel genes. But if you want only novel genes, then you can align reads to Genome outside of the reference GTF file and then perform cufflinks and cuffmerge without reference GTF and cuffdiff with merged.gtf.

[Lindsay44]

Is it necessary to include the gene annotation in Hisat2 (I am not using Tophat) in addition to the indexed reference genome? What are the pros of doing this?

[Satyajeet Khare]

In tophat, it helped me avoid generation of multiple XLOC ids for single gene id. So I use GTF for tophat. But this happens rarely and only a handful of genes get multiple XLOC ids. I dont see any disadvantage. Dont know about HISAT2. But if HISAT2 allows reference GTF, I will use it.

[Lindsay44]

Okay, thank you. What is the benefit of including the reference GTF in Cuffdiff as well?

[Satyajeet Khare]

GTF file is essential for cuffdiff as it will act as a template to quantitate transcripts in GTF file using aligned reads from bam file.

[Lindsay44]

can it be the merged gtf created from cuffmerge or does the reference have to be included as well?

[Satyajeet Khare]

Use either merged.gtf or reference GTF, not both.