How to run the Deseq2 tool without replicates.

Question

[fatimarasool135]

Hi,

I am doing differential gene expression by Deseq2 tool. I have no replicates and I am doing multi group comparison between samples.

Can anyone tell me how to run the Deseq2 tool without replicates?

[ATpoint]

Some links to previous topics towards that:

DESeq2 (or EdgeR) Exploratory Analysis with no Replicates

https://support.bioconductor.org/p/101210/

https://support.bioconductor.org/p/77623/

http://seqanswers.com/forums/showthread.php?t=31036

Deseq Analysis With Two Samples Without Replicates, Most Padj Equal To 1 And Na

[fatimarasool135]

Yes i have no replicates in any group and i nothing found answer to my question in manual of Deseq

[ATpoint]

From the manual (Nicolas Rosewick even linked it):

Can I use DESeq2 to analyze a dataset without replicates?

If a DESeqDataSet is provided with an experimental design without replicates, a warning is printed, that the samples are treated as replicates for estimation of dispersion. This kind of analysis is only useful for exploring the data, but will not provide the kind of proper statistical inference on differences between groups. Without biological replicates, it is not possible to estimate the biological variability of each gene. More details can be found in the manual page for ?DESeq.

The answers to your question are availablevia a solid search in the web, this literally has been discussed like 100 times before. Please show some effort and look for these threads.

[waqaskhokhar999]

Its better to use EdgeR for data without replicates. Please see the section 2.11 (What to do if you have no replicates)

[ATpoint]

No, it is not. If you read this section you refer to you will see that

We do not recommend any of these choices (edit: of the strategies suggested in the section) as a satisfactory alternative for biological replication.

Any framework, be it edgeR or DESeq2 that relies on dispersion estimation between replicates does not give reasonable results in the absence of replication.

It is very simple: If you want reliable statistics, do replicates. If you have none (be it due to money, material availability/feasibility or poor experimental design) the best you can do is to select some candidate genes based on ranking FCs (that have reasonable counts to avoid high FCs based on low counts) and validate them using independent experiments. You are never going to make strong claims on unreplicated data. ...and yes, there are even highly-cited publications that make strong claims on unreplicated data, and one should wonder how many of these claims really held true if one repeated things using proper design and statistics.

[lieven.sterck]

Hi waqaskhokhar999 ,

thanks for your input (which is indeed "correct" and I assume the other people answering here are aware of that).

However, the point we want to make is to "educate" people that they should not do DEG analyses without replicates. This point is as well mentioned in the EdgeR section you direct to. It anyway remains a 'hack' approach.

[chongyi.ng17]

The issue I have is that people like to pool samples together when uploading them. For example: https://www.ebi.ac.uk/ena/browser/view/SRR10857342 this contains rna seq data from several timepoints pooled together. I have no idea how to separate them or how to run deseq2 without 'replicates'.

[ATpoint]

Please open a new question, try to add as much detail as you can.

Answer 1

[Nicolas Rosewick]

None of your groups has replicates ? If not it will be very difficult (even impossible) to have reliable results ..

Nevertheless in the DESeq2 vignette : https://bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#can-i-use-deseq2-to-analyze-a-dataset-without-replicates

[fatimarasool135]

Yes i have no replicates in any group and i nothing found answer to my question in manual of Deseq

[Nicolas Rosewick]

As stated in DESeq2 manual :

This kind of analysis is only useful for exploring the data, but will not provide the kind of proper statistical inference on differences between groups. Without biological replicates, it is not possible to estimate the biological variability of each gene. More details can be found in the manual page for ?DESeq.

Answer 2

[lieven.sterck]

I'm afraid the only valid answer here will be NOT .

if you insist have a look here: http://bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html (first hit on google btw)