I have a question for deg analysis tools

Question

[likeabboo]

Hi.

I'm going to do DEG analysis with tmp data that has already been normalized.

I want to use a total of four tools; DESeq2, edgeR, Ballgown, Limma.

But I already knew the raw count data can only be used in DESeq2 and edgeR. And Limma can use tpm data(limma-trend method). Also Ballgown can use fpkm data unless using tablemaker or Stringtie.

Is this the correct content?

Answer 1

[ATpoint]

edgeR and DESeq2 assume raw counts for which no correction for gene length has been applied so take those out.

ballgown is for transcript rather than gene level analysis, mixing these two concepts up is a common mistake. In fact the entire hisat-stringtie-ballgown pipeline has never been made for DEG analysis and (even by the title of the Nature Protocols paper) is mainly for differential transcript analysis so take it out.

What is left is limma which might be the best choice with normalized data, see https://support.bioconductor.org/p/56275/#56299

My usual word of caution: If you don't have raw counts or raw data to generate those then your analysis is strictly speaking not reproducible and by this imo not publishable.

[likeabboo]

Thanks for your answer! :)

Additionaly, I have a one question more... I found a 'ballgownrsem: load RSEM data into a ballgown object' in R package. Do you know it??

If I use it, input data can be rsem normalized data? (not using tablemaker or stringtie)

I'm sorry about a little bit weird my question...

https://rdrr.io/github/alyssafrazee/ballgown/man/ballgownrsem.html

[ATpoint]

rsem does not normalize data, it's another approach to get raw counts/abundances. I do not know the function. ballgown is still a transcript not gene level tool.