Hi, dear friends, I got a strange thing when doing DESeq2 analysis, maybe I just made a silly mistake. Briefly speaking, sample B is much higher than sample A and C, all data are normalized by DESeq2, yet DESeq2 missed this significant group. Could anyone please tell me why? Thank you very much!
My command lines are as follows:
library(DESeq2) sample_test<-read.table("sample_test.txt",header=T) counts_test<-read.table("counts_test.txt",header=T,row.names=1) dds <- DESeqDataSetFromMatrix( countData=counts_test[rowSums(counts_test)>0,], colData=sample_test, design=~ condition) dds<-DESeq(dds) library("apeglm") res<- lfcShrink(dds, coef="condition_B_vs_A", type="apeglm") write.table(res,"DESeq2_result.txt",sep="\t",quote=F) norm<-counts(dds,normalized=T) write.table(norm,"test_normalize.txt",sep="\t",quote=F)
The strange thing happens in Chr6_6392513_6392788_peak. In normalized data, A sample(3 rep), B sample(3 rep) and C sample(3 rep):
Chr6_6392513_6392788_peak 178.648023305415 1.12212517818275 5.79671671533368 1246.9014826137 1124.82594062888 778.761332712122 0 3.66331655600197 0.
Obviously, B is much higher than A and C.
DESeq2_result.txt, which compares B and A, I got the following result:
Chr6_6392513_6392788_peak 371.079881967738 0.0246548548440441 0.1784530598261 0.026487239202988 0.347601564343674.
The padj is 0.35, not significant when padj threshold is set as 0.1.
The files are:
Thank you for your attention!