This is a test version of Biostars. For the public version, visit https://www.biostars.org.
error while using edgeR( Negative counts not allowed)

I am doing an RNAseq analysis. (using edgeR) I am loading my data for differential analysis, my data is of normalized gene expression(FPKM). I am now just trying to make DEGList object but, it is showing "Negative counts not allowed".

I have removed all negative values from the data, it is still showing the error(( Negative counts not allowed).

Please suggest what should i do. thank you!!

Command:

dgeFull <- DGEList(y, group = sampleInfo$condition)
Error: Negative counts not allowed
rna-seq r edger
• 5,842 views • link
updated by zx8754 zx8754 • written by deeksha3696 • deeksha3696

I have removed all negative values from the data,

How did you do that? Could you provide your count matrix?

EDIT: Forget this. Follow @Wouter comment.

• 0 views • link
written by Santosh Anand Santosh Anand

1 answer

my data is of normalized gene expression(FPKM).

You need raw counts for edgeR. Don't use any normalization.

• 1 views • link
written by WouterDeCoster WouterDeCoster

thanks from where can i get the raw counts

• 1 views • link
written by deeksha3696 • deeksha3696

From where did you get the fpkm counts...?

We are bad at reading your mind so you'll have to elaborate.

• 1 views • link
written by WouterDeCoster WouterDeCoster

I downloaded the fpkm values from NCBI GEO DATA SET

• 1 views • link
written by deeksha3696 • deeksha3696

Please add a link to the dataset you are using. It may be necessary to download the reads and process those yourself.

• 1 views • link
written by WouterDeCoster WouterDeCoster

FPKM values are not suitable for statistics, this is a recurrent discussion. See also edgeR user's guide, end of 2.3 on page 12. You need raw read counts, if they don't provide them at GEO, you can redo the analysis: download fastq files -> align them to reference e.g. with STAR -> counts reads with e.g. featureCounts

• 1 views • link
written by Benn Benn

While I don't completely agree that FPKM can't be used for any statistics (and I think it can be useful for visualization), it is true that you need raw counts for edgeR (and most RNA-Seq Bioconductor differential expression packages).

I also agree re-processing raw data can be frequently be helpful. Even if you end up with a similar result, you should understand the analysis process better and be more confident in your results :)

• 1 views • link
written by Charles Warden Charles Warden

Log in to answer this question.

More posts like this