MetaVolcanoR

Meta-analysis visualization tool with publication-ready customization

Overview

MetaVolcanoR combines differential gene, transcript, methylation and protein expression results from multiple studies to identify consistently perturbed genes. It implements three complementary meta-analysis strategies:

1. Random Effects Model (REM) - Rigorous statistical meta-analysis accounting for study variance
2. Vote-counting - Quick exploration of cross-study DEG consistency
3. Combining approach - P-value aggregation using Fisher's method

All methods exploit volcano plot reasoning for intuitive visualization of meta-analysis results.

Installation

From GitHub (Latest Development Version)

# Install devtools if needed
install.packages("devtools")

# Install MetaVolcanoR
devtools::install_github("iza-mcac/MetaVolcanoR")

From Bioconductor (Coming Soon)

# Will be available after Bioconductor submission
BiocManager::install("MetaVolcanoR")

Quick Start

library(MetaVolcanoR)

# Load example data (5 studies, ~20k genes each)
data(diffexplist)

# Run Random Effects Model meta-analysis
meta_results <- rem_mv(
  diffexp = diffexplist,
  metathr = 0.01,
  outputfolder = tempdir(),
  draw = "HTML"
)

# View interactive volcano plot
meta_results@MetaVolcano

# Explore forest plot for specific gene
draw_forest(meta_results, gene = "MMP9", draw = "PDF")

New Features 🎨

Plot Customization:

• Custom color schemes (including colorblind-friendly palettes)
• Automatic or manual gene labeling
• Adjustable point sizes and plot dimensions
• Custom titles and legends

Data Preparation Helpers:

• prepare_deseq2() - One-line conversion from DESeq2 results
• prepare_limma() - Format limma/voom output
• prepare_edger() - Convert edgeR results
• prepare_swish() - Convert swish/Fishpond transcript-level results
• prepare_limma() - One-line conversion from limma/voom results
• prepare_edger() - One-line conversion from edgeR results

Example with customization:

meta_custom <- rem_mv(
  diffexp = diffexplist,
  metathr = 0.01,
  # Customization options:
  colors = c(low = "navy", mid = "white", high = "darkred", na = "gray80"),
  label_genes = c("MMP9", "COL6A6", "MXRA5"),  # Specific genes
  label_size = 4,
  plot_title = "Disease vs Control Meta-Analysis",
  show_legend = TRUE
)

Documentation

📖 Full Tutorial and Examples - Comprehensive vignette with:

• Data preparation from DESeq2/limma/edgeR
• All three meta-analysis methods explained
• Customization gallery
• Publication tips

Input Data Requirements

Provide a named list of data frames, each containing:

• Gene identifiers (names or IDs)
• Log2 fold changes
• P-values
• Optional: Confidence intervals or variance (required for REM)

Quick data prep from DESeq2:

library(DESeq2)
dds <- DESeq(dds)
res <- results(dds)

# One-line conversion!
deg_table <- prepare_deseq2(res)

Three Meta-Analysis Methods

1. Random Effects Model (REM)

Most rigorous approach - accounts for between-study heterogeneity.

rem_results <- rem_mv(diffexp = study_list, metathr = 0.01)

When to use: You have confidence intervals or standard errors

2. Vote-Counting

Fast exploration of cross-study DEG consistency.

vote_results <- votecount_mv(
  diffexp = study_list,
  pvalue = 0.05,
  foldchange = 0.5
)

When to use: Quick overview, studies use different platforms

3. Combining Approach

Aggregates p-values (Fisher's method) and averages fold changes.

comb_results <- combining_mv(
  diffexp = study_list,
  metafc = "Mean"  # or "Median"
)

When to use: Focus on statistical evidence aggregation

License

GPL-3

---

Contributors: Izabela Mamede, Bárbara Marinho, Cesar Prada, Diogenes Lima, Helder Nakaya
Maintainer (at the moment): Izabela Mamede (iza.mamede@gmail.com)