biological replicates cuffmerge?

[KateShepard]

Dear forum,

I have 5 RNA-seq biological replicates and wish to examine the expression levels of specific genes/pathways of interest (high/low/not expressed) (no differential expression analysis).

I have mapped all samples to the appropriate ref genome/annotation using tophat2 and have assembled transcriptomes using cufflinks.

My question is now should I merge the 5 cufflinks assemblies together using cuffmerge, and use the merged assembly as my master transcriptome?

Also, is there a good tool/approach to visualise the variability between my biological replicates?

Thank you for your advice
Kate

[fanli]

My question is now should I merge the 5 cufflinks assemblies together using cuffmerge, and use the merged assembly as my master transcriptome?

Yes, that's the intended purpose of cuffmerge.

Also, is there a good tool/approach to visualise the variability between my biological replicates?

PCA, heatmaps, and clustering are useful for examining variability in your replicates. Check out the DESeq2 or edgeR packages for details.

[Michael.Ante]

In addition to fanli's post, you should have a look at the cummeRbund package for R. It was designed by the Cufflinks group and has some tailored analysis/visualisations for Cufflinks results.

Cheers,
Michael

[KateShepard]

CummeRbund for one condition (multiple biological replicates)

Thank you for your replies.

CummeRbund looks extremely useful, however I am having trouble working out how to use Cufflink data which originates from one condition only (multiple biological replicates). It seems like you can only use the output from cuffdiff which compares two conditions.

Do you know an approach for visualising data from one condition?

Thanks,
Kate