Home

dnaEPICO

dnaEPICO is an R/Bioconductor package for preprocessing and statistically analysing Illumina DNA methylation array data.

The package supports:

HumanMethylationEPIC v2.0 arrays
HumanMethylationEPIC arrays
HumanMethylation450K arrays
Local workstation and HPC execution
Automated workflows using GNU Make
Interactive Quarto reports

Overview

dnaEPICO provides a structured workflow that connects the main stages of an epigenome-wide association study:

Import raw red and green IDAT files.
Perform sample and probe quality control.
Predict sample sex and optionally remove confirmed sex mismatches.
Normalise methylation measurements.
Filter unreliable, SNP-associated, cross-reactive, and selected manifest-flagged probes.
Extract beta values, M-values, and copy-number measurements.
Estimate cell composition.
Estimate control-probe surrogate variables.
Prepare phenotype and methylation data for individual and combined timepoints.
Fit CpG-wise GLM or longitudinal mixed-effects models.
Annotate model results with genomic information.
Generate workbooks, logs, figures, result files, and an interactive web report.

Statistical models

Generalised linear models

CpG-wise GLMs are fitted using glm2.

The GLM workflow supports:

Numeric, binary, and multilevel phenotypes
Multiple phenotypes
Covariate adjustment
Factor variables
Scaling of selected numeric variables
Optional interaction terms
Multiple-testing adjustment
Model messages for warnings and fitting errors

Longitudinal mixed-effects models

Longitudinal models can be fitted using:

lmerTest and lme4
nlme

The default participant structure is a random intercept:

(1 | Participant)

The longitudinal workflow supports:

Repeated measurements
Phenotype-by-time interactions
Multiple phenotypes
Covariate adjustment
Scaling of selected numeric variables
Optional nlme residual correlation structures
Model messages for warnings and fitting errors

For lmerTest/lme4, an optional omnibus F test can provide one joint p-value for a complete multilevel phenotype or phenotype-by-interaction term. Satterthwaite and Kenward–Roger denominator degrees-of-freedom methods are supported.

Performance and restart support

CpG models are processed in bounded response blocks. Complete native model objects are not retained for every CpG.

The number supplied through N_CORES is treated as a maximum. dnaEPICO can select fewer workers according to:

Available CPUs
Available memory
Scheduler allocations
Number of CpGs
Selected model engine

After a phenotype is completed, dnaEPICO saves a compact phenotype summary. When:

RESUME_FROM_SUMMARY = TRUE

a compatible completed phenotype can be reused without fitting all its CpGs again.

An interrupted phenotype without a complete summary restarts from its first CpG.

Quick installation

Install the released Bioconductor version:

if (!requireNamespace("BiocManager", quietly = TRUE)) {
    install.packages("BiocManager")
}

BiocManager::install("dnaEPICO")

Install the development version from GitHub:

if (!requireNamespace("BiocManager", quietly = TRUE)) {
    install.packages("BiocManager")
}

BiocManager::install(
    "paulYRP/dnaEPICO",
    ask = FALSE,
    update = FALSE
)

Confirm the installation:

library(dnaEPICO)

packageVersion("dnaEPICO")

Quick start

Create a project directory and place the shared input files under:

data/preprocessingMinfiEwasWater/

The directory should contain:

data/preprocessingMinfiEwasWater/
├── idats/
├── pheno.csv
└── optional probe-exclusion files

Export the current Makefile template:

library(dnaEPICO)

extractMake(
    destDir = "/path/to/project",
    overwrite = FALSE
)

After configuring the Makefile, run the complete pipeline:

make all MODEL=model1

Other available routes include:

make f3 MODEL=model1
make f4 MODEL=model1
make f3lme MODEL=model1
make status MODEL=model1
make clean MODEL=model1

Main outputs

The workflow generates model-specific outputs under:

data/model1/
rData/model1/
results/model1/
figures/model1/
logs/model1/
reports/model1/

The principal model workbooks are:

data/model1/methylationGLM/annotatedGLM.xlsx
data/model1/methylationLME/annotatedLME.xlsx

Each workbook contains:

Annotated model results
Analysis metadata
A column dictionary

The interactive report is generated at:

reports/model1/docs/index.html

The complete reports/model1 directory must remain together because the HTML page uses supporting result files stored under reports/model1/assets/.

Wiki navigation

Additional resources

Citation

Display the current citation from R:

citation("dnaEPICO")

Please also cite the underlying Bioconductor and statistical packages used in the selected workflow.

Support

Questions about package usage can be submitted through:

Navigation Menu