DE Results - Usage Guide

Overview

This skill covers extracting, filtering, annotating, and exporting differential expression results for downstream analysis. Works with both DESeq2 and edgeR output.

Prerequisites

# For annotation
BiocManager::install(c('org.Hs.eg.db', 'biomaRt'))
# For Excel export
install.packages('openxlsx')

Quick Start

Tell your AI agent what you want to do: - "Extract significant genes with padj < 0.05 and |log2FC| > 1" - "Add gene symbols to my DESeq2 results" - "Export results to Excel with separate sheets for up and down genes"

Example Prompts

Filtering

"Get the top 100 most significant genes"

"Filter for up-regulated genes with at least 2-fold change"

"Find genes significant in both DESeq2 and edgeR"

Annotation

"Add gene symbols and descriptions from Ensembl"

"Convert Ensembl IDs to gene symbols"

"Annotate results with my custom annotation file"

Export

"Save significant genes to CSV"

"Create an Excel file with all results and significant genes"

"Prepare a ranked gene list for GSEA"

What the Agent Will Do

Extract results from DESeq2/edgeR object
Apply requested filters (padj, log2FC, baseMean)
Add gene annotations if requested
Format and export results

Common Thresholds

Filter	Typical Value	Description
padj	< 0.05	Standard significance
padj	< 0.01	Stringent
\|log2FC\|	> 1	2-fold change
\|log2FC\|	> 0.585	1.5-fold change
baseMean	> 10	Minimum expression

Tips

Always use adjusted p-values (padj/FDR), not raw p-values
Check for NA values in padj -- three distinct causes: zero counts, Cook's distance outliers, independent filtering threshold
Use lfcShrink() results for ranking genes and GSEA input
Export both all results and filtered significant genes
For GSEA: rank ALL genes (no cutoff); for ORA: use significant genes + background
Check p-value histogram before trusting results (uniform + spike at 0 is correct)
For prokaryotic organisms, use Prokka/Bakta GFF for annotation and strain-specific KEGG codes

deseq2-basics - Run DESeq2 analysis
edger-basics - Run edgeR analysis
de-visualization - Visualize results
pathway-analysis/go-enrichment - GO over-representation analysis
pathway-analysis/gsea - Gene set enrichment analysis