Question

ControlGenes/ housekeeping genes Deseq2

4

Entering edit mode

mankadeep2 ▴ 40

@mankadeep2-23453

Last seen 4.5 years ago

I had a query regarding what DESeq2 assigns as ControlGenes (housekeeping genes). As per the manual, the Normalized Counts can be determined by estimateSizeFactorsForMatrix(counts, locfunc = stats::median, geoMeans, controlGenes) While, geoMean is explained, controlGenes are the first 200 genes? Could anyone please explain the criteria for selecting the first 200 genes as Control? Is there a way in DESeq2 which can be used to determine what are the housekeeping genes and plot their expression?

normalization deseq2 sizefactors • 4.0k views

ADD COMMENT • link updated 4.6 years ago by Michael Love 43k • written 4.6 years ago by mankadeep2 ▴ 40

score 0 · Answer 1 · 2020-04-30

0

Entering edit mode

Michael Love 43k

@mikelove

Last seen 2 days ago

United States

controlGenes are specified by the user, as to which genes to use for calculating the size factor. By default all genes are used.

ADD COMMENT • link 4.6 years ago Michael Love 43k

0

Entering edit mode

Thanks for the reply, just few follow up questions. Will adding controlGenes also result in change in output of (by itself) dds <- DESeq(dds) or it would have to be defined herein?

ADD REPLY • link 4.6 years ago mankadeep2 ▴ 40

0

Entering edit mode

If you run estimateSizeFactors before DESeq it will use those pre-estimated size factors, and it will print a message saying so.

ADD REPLY • link 4.6 years ago Michael Love 43k

0

Entering edit mode

So, I had uploaded genes to controlGenes <- c("BJA_RS02215", "BJA_RS03430", "BJA_RS04155", "BJA_RS05410", "BJA_RS07010"), but this creates a character class object, which when used to run

dds <- estimateSizeFactors(dds, type = c("ratio","poscounts", "iterate"), locfunc = stats::median, geoMeans,controlGenes)

throws out a very expected error:

Error in estimateSizeFactorsForMatrix(counts(object), locfunc = locfunc,  : 
  controlGenes should be either a numeric or logical vector

I have been trying to convert it to either of them using as.is and have failed to accomplish it. Is there another way of converting gene names into a numeric/logic object?

ADD REPLY • link 4.6 years ago mankadeep2 ▴ 40

0

Entering edit mode

Whenever trying out something new, you should check the help for the function.

?estimateSizeFactors

will tell you that controlGenes is:

optional, numeric or logical index vector specifying those genes to use for size factor estimation (e.g. housekeeping or spike-in genes)

A logical vector in R is a vector of TRUE or FALSE that can be used to index another vector or matrix-like object. Here you can just do:

isControl <- rownames(dds) %in% <your.gene.names.go.here>