Question: Multiple group analysis
0
3.8 years ago by
John30
John30 wrote:

Hi,

Is it possible to do multiple group level analysis?

T1(6replication)

T2(6replication)

T3(6replication)

T4(6replication)

I tried it but I could only see T6 vs T1. How can I check the other comparisons?

{r DEseq2_analysis_t1_ld, include=FALSE}
ddsHTSeq<-DESeqDataSetFromHTSeqCount(sampleTable=sampleTable, directory=INDIR, design=~treatment)
dds<-DESeq(ddsHTSeq)
ddsNozero  <-  which(!mcols(dds)$allZero) dds <- dds[ddsNozero] res<-results(dds) res<-res[order(res$padj),]


> res
log2 fold change (MAP): treatment T6 vs T1
Wald test p-value: treatment T6 vs T1
DataFrame with 49044 rows and 6 columns

Thank you for the help.

J.

deseq2 • 1.2k views
modified 3.8 years ago • written 3.8 years ago by John30
2
3.8 years ago by
tony.fox201630
tony.fox201630 wrote:

res2<-results(dds, contrast=c("treatment","T1", "T2"), test="Wald")

res3<-results(dds, contrast=c("treatment","T1", "T3"), test="Wald")

...

1

Yes, and in this case, you could leave off the test="Wald" (because DESeq() was run without setting test="LRT")

What does LRT stand for??

The fastest way to learn about functions and there arguments is to check the manual pages for each function. For example:

?DESeq

or

?results

has lots of detailed information, and on top of that you can check the vignette:

vignette("DESeq2")

Thank you! It means likelihood ratio test!

LRT= Likelihood Ratio Test

0
3.8 years ago by
John30
John30 wrote:

Thank you so much.

@Michael, is that because test="Wald" is the default?