How to identify the Up and down Regulated genes from Cuff diff output

[sridhar28]

Dear All,

How to identify the Up and down regulated genes from the diff genes, I got diffgenes using Cummerbund codes.

In my diff_gene.txt file could see most of the log2fold change values are +inf and -inf. From the list how to separate the up and down regulated genes??

Cheers

Sridhar

[bassounds]

Hey Sridhar,

I am also getting similar results (+inf and -inf) on some of the log2fold change values. I believe this is happening because one of the conditions has a value of 0.

I am reading through cuffdiff documentation to sort out whether this is due to normalization or processing issue, or whether I actually have zero reads for some transcripts. I doubt the latter is the case, based on my data of identical cell lines with different treatments, and high coverage of the ran-seq reads.

[velt]

Hello,

In my view, -inf and inf values, are effectively due to a division by 0 when calculating the fold-change.

For example in a test_vs_control comparison, inf value for a gene means that there is many reads in test condition and 0 in the control. And conversely, -inf value for a gene means that there is 0 read for this gene in test condition and many in control condition.

Do not ignore these genes, because if the p-value is significant, they can be considered deregulated.

For me, there are two solutions, a "clean solution" and "less clean solution".

- clean solution => replace the 0 values by a non-zero value (1 for example) and recalculate the fold-change for these genes.

- less clean solution, but simpler => if the exact value of the fold-change is not important to you, and you only use it to filter the differentially expressed genes, just replace the inf & -inf values by a large value (-1000 and 1000 for example). Thus, you can select those genes that have a high fold-change (if the p-value is significant of course).

Otherwise, to select genes differentially expressed, you can filter diff_gene.txt file according to two criteria, the log2-fold change and p-value. Generally a gene is considered significant if its p-value is less than 0.05, but it is up to you to set a threshold. In addition, you can consider that a gene is truly differentially expressed if its log2 fold-change is > 1 or < -1. But again, it's up to you to set that threshold.

Thus, down-regulated genes have a fold-change <= -1 and a p-value <= 0.05 and up-regulated genes have a fold-change >= 1 and a p-value <= 0.05.

I hope these ideas will help you.
(PS: sorry if my english is bad, this language is not my native language)