DECODeR
Differential Exploration of Counts-based Omics Data in R
Where Do I Start?
Already have DEG results?
Upload your differential expression tables, counts, and metadata
Need to run differential analysis?
Start with raw counts and perform differential expression analysis
What can I do with DECODeR?
Perform differential expression analysis of count data or import a DEG table you already have.
Filter and explore differentially expressed genes with dynamic tables
Create volcano plots, heatmaps, PCA plots, and dimension reduction visualizations
Perform GO enrichment and KEGG pathway analysis on your gene sets
About DECODeR
DECODeR (Differential exploration of counts-based omics data in R) is an interactive toolkit designed by the University of Florida Health Cancer Center BCB-SR Bioinformatics unit for comprehensive analysis and visualization of RNA-seq data. Upload your DEG results or perform differential expression analysis on your counts to create publication-ready plots, explore differential expression patterns, and conduct downstream analyses—all through an intuitive web interface.
Getting Help
Questions about this app or your analysis? Contact UFHCC-BCBSR@ufl.edu
Load Your Data
Upload your files to begin analysis and visualization
Step 1: Choose Your Data Type
Normalized Counts + DEG Table
I have normalized data and differential expression results
Raw Counts Only
I need to perform differential expression analysis
Step 2: Upload Files
Try DECODeR with Demo Data
Load example RNA-seq data instantly
Normalized Count Matrix
Sample Metadata
DEG Table
Download Templates:
Step 3: Preview Your Data
Ready for Differential Analysis!
Since you have raw counts, you can perform differential expression analysis directly in DECODeR.
The results will automatically be available to all visualization modules - no need to upload anything here!
Differential Expression Analysis
Guided statistical analysis for RNA-seq count data
Volcano Plot Analysis
Interactive visualization of differential gene expression with significance thresholds
Plot Settings
Contrast Selection
Significance Thresholds
Gene Labeling
Volcano Plot
Dimensionality Reduction
Comprehensive analysis with PCA, UMAP, and sample correlation assessment
Analysis Settings
Group Selection
Gene Selection
Expression Filtering
UMAP Parameters
Analysis Overview
PCA: Linear dimensionality reduction showing major variance patterns.
UMAP: Non-linear reduction revealing local clustering structure.
Correlation: Sample-to-sample similarity matrix for quality assessment.
Variance filtering: Uses most variable genes rather than differential expression.
Dimensionality Reduction Analysis
Principal Component Analysis
UMAP Projection
Sample Correlation Matrix
Interpretation Guide:
• UMAP: Preserves local neighborhoods, reveals clusters
• Variance genes: More stable than DEG-based analysis
Interactive Heatmaps
Visualize gene expression patterns across samples and conditions
Filter Settings
Statistical Thresholds
Gene Selection
Custom Gene Upload:
Upload a CSV or TSV file with a 'SYMBOL' column containing gene names.
Try with Demo Genes
Load example gene list instantly
Contrast Selection
Color Schemes
About Heatmaps
Row scaling: Gene expression values are standardized (z-score) across samples to highlight relative changes.
Clustering: Hierarchical clustering groups similar genes (rows) and samples (columns).
Colors: Intensity represents expression relative to each gene's average across all samples.
Expression Heatmap
Generating heatmap...
Differential Expression Results
Browse, filter, and export differentially expressed genes with statistical summaries
Filter Settings
Contrast Selection
Statistical Thresholds
Summary Statistics
Distribution Overview
Table Features
Sorting: Click column headers to sort data.
Search: Use column filters to find specific genes.
Export: Use table buttons to export results.
Colors: Red = up-regulated, Blue = down-regulated.
Differential Expression Table
Export Options: Use the buttons above the table to download results in various formats (CSV, Excel, PDF).
Enrichment Analysis
Gene Ontology and KEGG pathway enrichment analysis for differential expression results
Select Analysis Type
Analysis Settings
Contrast Selection
DEG Selection
Enrichment Parameters
GO Specific Options
About GO Analysis
Gene Ontology: Systematic classification of gene functions across three domains.
Enrichment: Statistical overrepresentation of GO terms in your gene set.
Visualization: Multiple plot types reveal different aspects of enrichment patterns.
Interpretation: Lower p-values and higher gene ratios indicate stronger enrichment.
GO Enrichment Results
Analysis Settings
Contrast Selection
DEG Selection
Enrichment Parameters
Plotting Options
About KEGG Analysis
KEGG Pathways: Comprehensive database of biological pathways and molecular interactions.
Pathway Analysis: Identifies significantly enriched biological pathways in your gene set.
Pathway View: Interactive pathway diagrams show your genes mapped onto biological pathways.
Applications: Understand biological processes and mechanisms underlying your experimental results.
KEGG Pathway Results
About DECODeR
Differential Exploration of Counts-based Omics Data in R
Overview
DECODeR is an interactive toolkit designed by the University of Florida Health Cancer Center Bioinformatics and Computational Biology Shared Resource (BCB-SR) to support comprehensive analysis and visualization of RNA-seq data. The platform enables researchers to upload their own data or perform differential expression analysis directly within the application.
Frequently Asked Questions
Q: Should I upload normalized or raw count data?
A: It depends on your workflow. For differential expression analysis within DECODeR, upload raw counts. If you already have DEG results, upload normalized data (logCPM) along with your DEG table for visualization.
Q: Why do you use different data types for analysis vs. visualization?
A: Differential expression analysis requires raw counts to properly model the count-based nature of RNA-seq data. However, visualizations (PCA, heatmaps) work better with log-transformed, normalized data (logCPM) to reduce the influence of highly expressed genes.
Q: What file formats do you accept?
A: We accept CSV, TSV, and Excel files. Count matrices should have genes as rows and samples as columns. Metadata should include sample IDs that match your count matrix column names.
Q: Can I analyze complex experimental designs?
A: Currently, DECODeR supports two-group comparisons (multiple contrasts are fine) and designs with batch effects. Support for more complex designs (time series, factorial designs, etc.) is planned for future updates.
Q: How do I handle batch effects or unwanted variation?
A: DECODeR can detect and adjust for known batch effects when performing differential analysis. Advanced methods for removing unknown sources of variation (SVA, RUVseq) are planned for future releases.
Q: Which differential expression method should I use?
A: All three methods (DESeq2, limma-voom, edgeR) are robust. DESeq2 is generally recommended for smaller sample sizes, limma-voom for larger studies, and edgeR offers the fastest performance. Results are typically highly concordant.
Q: When will new features be available?
A: We continuously update DECODeR based on user feedback. Check back regularly for new features or contact us with specific requests.
Methods
This section describes the computational methods implemented in each DECODeR module:
Data Filtering
Low-abundance genes are filtered using edgeR's filterByExpr() function, which removes genes with consistently low counts across samples. This reduces multiple testing burden and improves statistical power (Robinson MD, et al. Bioinformatics 2010).
Batch Assessment
When batch information is provided, DECODeR assesses batch effects using principal component analysis and visualization. The user should then include the batch variable as a covariate when they configure their analysis. Currently we do not provide batch corrected counts for visualization purposes, but this is planned for a future release.
Differential Expression Analysis
- DESeq2: Uses negative binomial generalized linear models with empirical Bayes shrinkage for dispersion estimation (Love MI, et al. Genome Biology 2014).
- limma-voom: Applies precision weights to log-transformed counts, enabling linear modeling methods (Law CW, et al. Genome Biology 2014).
- edgeR: Employs empirical Bayes methods to moderate tagwise dispersions in negative binomial models (Robinson MD, et al. Bioinformatics 2010).
Visualizations
- PCA: Principal component analysis performed on log-transformed counts using prcomp() from base R.
- MDS: Multidimensional scaling using Euclidean distances with cmdscale() from base R.
- Volcano Plots: Created using ggplot2, displaying log2 fold change vs. -log10 adjusted p-values (Wickham H. ggplot2: Elegant Graphics for Data Analysis 2016).
- Heatmaps: Generated using pheatmap with hierarchical clustering and standardized gene expression values (Kolde R. pheatmap: Pretty Heatmaps 2019).
Enrichment Analysis
- Gene Ontology (GO): Over-representation analysis performed using clusterProfiler with hypergeometric testing and Benjamini-Hochberg correction (Yu G, et al. OMICS 2012).
- KEGG Pathways: Pathway enrichment analysis using clusterProfiler with KEGG.db annotation package (Carlson M. KEGG.db 2019).
Additional R Packages
DECODeR utilizes several additional R packages for data processing and visualization:
- Gene Annotation: org.Hs.eg.db, org.Mm.eg.db for mapping between gene identifiers (Carlson M. 2019)
- Data Manipulation: dplyr, tidyr for data wrangling (Wickham H, et al. 2023)
- Interactive Tables: DT for interactive data tables (Xie Y, et al. 2023)
- File I/O: readr, readxl for data import (Wickham H, et al. 2023)
- Shiny Framework: shiny, bslib for web application interface (Chang W, et al. 2023)
Support & Contact
DECODeR is developed and maintained by the UFHCC Bioinformatics and Computational Biology Shared Resource.
For technical support or questions: UFHCC-BCBSR@ufl.edu
For collaboration inquiries: Visit cancer.ufl.edu/research/shared-resources