I have a relatively large RNAseq data set: 360 Human blood biological replicates, 100bp PE reads with 60M read depth on average, with Poly-A removal.
I'm analyzing these data with DESeq2 following these basic steps:
dds = DESeqDataSetFromMatrix(countData = dat0, colData = colDat, design = ~ Age + Sex + RIN + pH + Treatment)
dds <- estimateSizeFactors(dds)
dds <- estimateDispersions(dds)
dds <- nbinomWaldTest(dds)
ddsRes <- results(dds, pAdjustMethod = "BH")
However, many of the most significant (including the top 6 results) are small ncRNA with relatively low baseMean counts (<10). Below I am attaching a plot of the p-values versus baseMean distribution, does this look abnormal? It certainly looks different than the ones presented in the DESeq2 tutorial found here: https://www.bioconductor.org/help/course-materials/2015/LearnBioconductorFeb2015/B02.1.1_RNASeqLab.html#diagnostic (under the Independent Filtering section).
•
written
by
dhibar •
1 answer
Is that equivalent to this?
dds <- DESeqDataSetFromMatrix(yourinput)
dds <- DESeq(dds)
res <- results(dds)
Alrighty then :p With regard to your results, I assume that since your enormous delicious sample size you're analysis is "overpowered" to find deferentially expressed genes. An approach to test this would be to perform the same analysis with far less samples in each group.
I think DESeq2 and others were developed in times when sequencing was far too expensive for this nice cohort.
Log in to answer this question.
-
Using the VST normalization rather than the DESeqDataSetFromMatrix function for DESeq2
written by pailloufat •
-
Outlier after DEseq2 in meta-analysis
written by a.stef.44 •
-
DESeq2 results function runs very slow on Windows 10
written by Taha •
-
Stringtie on non model organism - how to best proceed for DE analysis?
written by Shred
-
Normalize RiboTag samples to input before differential expression with DESeq2
written by lozzi •
-
Deseq2 weird values for Log2FC and pvalue.
written by biom.andressa •
-
DESeq2 4 group comparison
written by arabi •
-
RNA-Seq time series analysis using a DESeq2 spline approach yields far too many significant genes
written by stu111538
-
error in getting results from DESeq2
written by nazaninhoseinkhan
Consider posting/cross-linking this question on Bioconductor support site. @Michael Love monitors that more frequently (based on one of his recent posts here).