Different annotation gtf file in Tophat2 and Cufflink

Question

[aditi.qamra]

Hi,

I wanted to know if using a different reference annotation file in cufflinks from that used in Tophat would impact the results?

To explain a little more, tophat2 was run to get read alignments for rnaseq data using the -G option which basically tries and maps the reads according to the given reference file and then fills the remaining gaps. Now I want to run cufflinks downstream of it and use the RABT assembly however I do not have the same annotation gtf file that was used in Tophat ( tophat was run by somebody else and its been a really long time..)- I was wondering what sort of and how much of an impact would that make?

Thanks !

Aditi

Answer 1

[Vivek]

I suspect the impact won't be too much if its a model organism mainly because there aren't many new splice junctions detected in a short span of time. However if you intend to be sure you can always get the reads back from the BAM in fastq and do the alignment with the new GTF as the guide. Tophat doesn't trim read ends during mapping so you could get the original reads back.

Answer 2

[Istvan Albert]

The answer will of course depend on just how different are the two annotation files.

Answer 3

[aditi.qamra]

I suspect tophat used an ensembl gtf file maybe a 59 or 62 release and I was going to use the UCSC hg19 genes.gtf file from iGenome