how to identify differentially expressed genes from cuffdiff output?

[IBseq]

Hi guys,
have a question about the cuffdiff output "differential expression testing".

For most of you might sound "naive" but I'm new to this field and I
have very little background in statistic.

So, I have compared a control sample with 2 biological replicates
using cuffdiff. I have now about 4000 genes which were tested.

1. How do I extrapolate the genes which are up- or downregulated from the 4000?

2. Is there a FPKM value above which a gene is up- or downregulated?

3.I used excel and sorted the values from highest to smallest:
assuming that control highest value is 200 and the correspondent
treated values is 2, can I say that that gene is downregulated in the
treated ssamples by a 100 fold chnage?

4. Do I have to use at all the p_values given in the output to
extrapolate the most up- or downregualted genes?

I do not have yet cummerbund and I am not very good with R. And I 'm lost!

Thanks,
ib

[DZhang]

Hi IBseq,

Here are a few practical suggestions hopefully to get you started:

1) Decide whether you need transcript-level or gene-level quantitation. It depends on what kind of biological questions you are trying to address. (cuffdiff gives both, just in case you want to know.)
2) Please consider both fold change and p-value in deciding which genes/transcripts as most significantly changed for follow-up studies.
3) Always use independent experiments to verify the differentially expressed genes. Statistics has it own limitations, and so are the technologies.
4) For one specific question, yes, you may say that gene has 100-fold change. But be aware of the small numbers. The count of 2 is not very reliable. If the control has none and your experimental group has 1, mathematically, it is infinitely increase but in reality you cannot really tell.

Good luck learning R!

Douglas

https://www.contigexpress.com

[jp.]

May please someone guide me for pathway analysis after cuffdiff events?
I have some sig diff but I dont know how to do pathway analysis. Are there some commnads or rules something , a bit in detail, if somebody know?
thank you

[DonDolowy]

You can run GSEA to identify enriched gene sets in (here you submit a matrix containing expression values for all genes in between your two conditions).
If you want to study only the DEGs, then you can go for GO analysis.