What log2FoldChange Means in DESeq2 Output
A plain explanation of the log2FoldChange column in DESeq2 results, how the shrunken estimate differs from a ratio of means, and how to compute and check it yourself.
11 posts
A plain explanation of the log2FoldChange column in DESeq2 results, how the shrunken estimate differs from a ratio of means, and how to compute and check it yourself.
An MA plot puts fold change against mean expression; a volcano plot puts fold change against significance. Both are scatter plots of the same differential expression table, and each catches a different failure mode.
log2FC is the base-2 logarithm of a ratio between two expression estimates. This page explains the arithmetic, the shrinkage that DESeq2 and limma apply, and the failure modes that make a large log2FC meaningless.
A p-value is the error rate of one test under the null. An FDR of 0.05 is a property of the list you selected, meaning about 5% of the features you called significant are expected to be false. Here is how the two differ in practice, and how to compute them on your own omics data.
A working pipeline from FASTQ files to a ranked table of differentially expressed genes, including longitudinal sampling of one person.
A working guide to gene set enrichment analysis on personal RNA-seq: building a defensible ranked list with DESeq2, running fgsea against MSigDB, reading the enrichment plot, and using GSVA when you have no contrast.
A working pipeline from a counts matrix to a ranked, deduplicated pathway table for personal RNA-seq, including the n=1 designs that are valid, the blood cell composition problem, and which tools to trust.
A working transcriptome analysis workflow for one person's whole-blood RNA-seq. Covers QC, salmon quantification, normalization and pathways.
A working guide to setting up an R and Bioconductor project around a personal molecular profile: loading your VCF, quantifying your own RNA-seq across timepoints, normalizing proteomics, and joining biomarker and glucose time series into one analyzable object.
A working pipeline from salmon quantification files to a differential expression table and pathway results, written for one person's longitudinal blood transcriptome rather than a two-group lab experiment.
A working guide to running DESeq2 on bulk RNA-seq counts. Covers the count matrix, dispersion, shrinkage and the sanity checks that catch most mistakes.