multimethconv

Overview

multimethconv is an integrative pipeline for cell type deconvolution from methylation data. It reduces dimensionality and eliminates redundancy through unsupervised filtering and iterative correlation analyses. Built on top of existing frameworks like deconvME and multideconv, multimethconv harmonizes outputs across methods to identify robust cell subgroups and mitigate signature-driven heterogeneity.

Pipeline Schema

multimethconv Pipeline

Overview of the multimethconv analysis pipeline showing the integration of multiple deconvolution methods and consensus filtering approach.

Installation

You can install the development version of multimethconv from GitHub with:

# install.packages("pak")
pak::pak("VeraPancaldiLab/multimethconv")

Requires R (>= 4.3)

Core Functions

1. compute.deconvolution() - Main Deconvolution Function

Performs DNA methylation deconvolution on a beta matrix using multiple algorithms:

library(multimethconv)

# Run deconvolution with default methods
deconv_result <- compute.deconvolution(
  methyl_set = beta_matrix_example,
  methods = c("Epidish", "Methylresolver", "Methatlas", "Houseman", "MethylCC"),
  doParallel = FALSE,
  file_name = "my_analysis"
)

Supported Methods: - EpiDISH (RPC mode) - from deconvME - Methylresolver - from deconvME - MethAtlas - from deconvME

• Houseman - reference-based method - MethylCC - reference-free method

2. create_methylation_signature() - Generate Custom Signatures

Creates a DNA methylation signature by identifying differentially methylated positions (DMPs):

# Create custom methylation signature
signature <- create_methylation_signature(
  data = beta_matrix,
  Phenotype = phenotype_vector,
  FDR = 0.05,
  FCcuoff = 1.0,
  MaxDMRs = 100
)

3. compute_CpG_Signature_Extension() - Extend RNA Signatures

Extends RNA signatures with CpG-derived genes using PCHi-C network:

# Extend RNA signature with CpG information
extended_sig <- compute_CpG_signature_extension(
  rna_signature = rna_sig,
  cpg_signature = cpg_sig,
  cpg_region = "both"  # Options: "both", "promoter", "distal"
)

Features: - Maps differentially methylated CpGs to promoter and distal fragments - Uses PCHi-C network for chromatin interactions - Propagates CpG signals to associated genes - Combines promoter and distal gene signatures

4. cpg_to_gene_expr() - Convert CpG Scores to Gene-Level Expression

Maps CpG methylation scores to genes and converts them to pseudo-expression:

# Map CpG scores to genes
gene_expr <- cpg_to_gene_expr(cpg_gene_map, cpg_scores)

Key Features

• Multi-method Integration: Combines EpiDISH, MethylResolver, MethAtlas, Houseman, and MethylCC frameworks
• Flexible Reference Signatures: Supports custom methylation signatures or built-in references
• PCHi-C Network Integration: Leverages chromatin interaction networks for enhanced annotation
• Methylation to Expression: Converts CpG methylation patterns to gene-level pseudo-expression scores
• Comprehensive Cell Type Coverage: Identifies and harmonizes outputs for multiple immune and stromal cell types
• Dimensionality Reduction: Automated filtering of redundant cell types through correlation analysis
• Parallel Processing Support: Option to parallelize computationally intensive methods

Key Dependencies

The package integrates with:

• Deconvolution Frameworks: deconvMe (EpiDISH, MethylResolver, MethAtlas), multideconv
• Genomics: GenomicRanges, IRanges, S4Vectors, minfi
• Statistical Methods: limma, matrixStats
• Data Manipulation: dplyr, magrittr, stringr
• Network Analysis: igraph, chaser

Quick Start Example

library(multimethconv)

# Load or prepare your beta matrix
beta_matrix <- your_beta_values  # rows = CpGs, columns = samples

# Run full deconvolution pipeline
results <- compute.deconvolution(
  methyl_set = beta_matrix,
  methods = c("Epidish", "Houseman", "MethylCC"),
  file_name = "my_deconvolution"
)

# View results
head(results)

Cell Type Standardization

The package automatically harmonizes cell type nomenclature across methods, including:

• Macrophages (M0, M1, M2)
• Monocytes
• Neutrophils
• Natural Killer (NK) cells and NKT cells
• CD4+ T cells (naive, memory, regulatory)
• CD8+ T cells
• Helper T cells
• Gamma-delta T cells
• Dendritic cells (activated, resting)
• B cells (naive, memory)
• Cancer-associated fibroblasts (CAF)
• Endothelial cells
• And many more…

Vignettes

For detailed workflow examples and use cases:

browseVignettes("multimethconv")

Contributing

Contributions are welcome! Please submit a Pull Request or open an issue on GitHub.

License

This project is licensed under the GPL (>= 3) License - see the LICENSE file for details.

Authors

• Marcelo Hurtado (marcelo.hurtado@inserm.fr) - Creator and Maintainer
  • ORCID: 0009-0004-6712-0864
• Vera Pancaldi (vera.pancaldi@inserm.fr) - Co-author
  • ORCID: 0000-0002-7433-624X

Citation

If you use multimethconv in your research, please cite:

Hurtado, M., & Pancaldi, V. (2024). 
multimethconv: An integrative pipeline for cell type deconvolution from methylation data. 
R package version 0.0.1. 
https://github.com/VeraPancaldiLab/multimethconv