Question

DESeq2 rlog as input for GAGE

0

Entering edit mode

Tom ▴ 10

@tom-10763

Last seen 5.3 years ago

Dear all,

In the vignette of GAGE pathway analysis package, a joint workflow with DESeq2 is shown, where only the column of log2 fold change from DESeq2 output is used as an input for GAGE.

Is rlog matrix calculated by DESeq2 applicable as an input for GAGE pathway analysis?

Any comments would be appreciated.

Best,

Tom

deseq2 gage • 1.7k views

ADD COMMENT • link 8.5 years ago • updated 8.4 years ago Tom ▴ 10

score 0 · Answer 1 · 2016-07-13

0

Entering edit mode

Luo Weijun ★ 1.6k

@luo-weijun-1783

Last seen 18 months ago

United States

Not sure exact what rlog does in stabilizing variance. It may be safee to use just the normalized raw counts followed by log2 transformation, something like: cnts=count(dds, normalized=T) cnts.log=log2(cnts+1) cnts.log can then be used as gage input. Remember to specify ref and samp (column number vectors) though.

ADD COMMENT • link 8.5 years ago Luo Weijun ★ 1.6k

1

Entering edit mode

The idea of the vst() and rlog() functions in DESeq2 is to stabilize the variance of transformed counts across the range of values. These two functions are a motivated, statistical approach to transformation rather than picking either an arbitrary psuedocount and/or filter threshold for each dataset which will reduce the large noise component of log(count + x).

vst() and rlog() return data which is log2-like, but the variance has been stabilized for low counts, and should be appropriate for downstream usage.

ADD REPLY • link 8.4 years ago Michael Love 43k

score 0 · Answer 2 · 2016-07-14

0

Entering edit mode

Tom ▴ 10

@tom-10763

Last seen 5.3 years ago

Luo and Michael,

Thank you for your helpful comments. I will use rlog transformed data as an input for downstream GAGE analysis.

Best,

Tom

ADD COMMENT • link 8.4 years ago Tom ▴ 10