Computational genomics · machine learning

From genomic data to biological discovery.

We develop and apply computational frameworks, statistical learning and machine-learning methods to interrogate genome regulation, cellular heterogeneity and cancer biology. Our research translates high-dimensional multiomics data into biologically interpretable findings.

Featured software & methods

Methods grounded in published research.

3D GENOME · WORKFLOW

HiC-ECC

A modular framework for enhancing and comparing Hi-C matrices and characterizing chromatin interactions and genome architecture. Integrates established deep-learning-based enhancement methods; the lab did not create those underlying networks.

HiC-ECC is an ongoing computational framework; see the linked meeting abstract.

AACR 2025 abstract

CHROMATIN STATES · COMPARATIVE ANALYSIS

ChromXploreR

A collaborative computational tool for comparison and interpretation of chromatin-state landscapes in different sample groups.

Developed through computational research collaborations.

GitHub repository

STATISTICAL LEARNING · REGRESSION

Methylation-eQTL

A sequential penalized-regression method for linking CpG methylation profiles and gene-expression variation across cancer datasets.

Peer-reviewed statistical methodology.

Bioinformatics paper

REPRODUCIBLE EPIGENOMICS · METHODS

CUT&RUN benchmarking

A published head-to-head evaluation of MACS2, SEACR, GoPeaks and LanceOtron across multiple CUT&RUN histone-mark datasets.

Published in Bioinformatics (2025); benchmarking includes evaluation of a machine-learning-based caller.

Public analysis code

Machine learning & AI research

Learning from complex biological data.

Deep-learning-assisted 3D genomics

Using established deep learning for contact-map enhancement within reproducible 3D genome analysis pipelines.

Integrating established neural-network approaches in genomic analysis.

Unsupervised learning for single-cell multiomics

Integrating chromatin accessibility and gene expression with dimensionality reduction, clustering and regulatory program modelling.

Used in integrative epigenomics and single-cell research.

Predictive epigenomics & computational oncology

Exploring multimodal modelling, candidate biomarker discovery and the relationship between molecular features and therapeutic response.

Research applications in multimodal cancer genomics.

Open data & community resources

Reusable evidence. Reproducible science.

Single-cell chromatin accessibility atlas

51,248 cells profiled across nine mouse tissues. Public accession GSE286329.

Explore GEO data

CUT&RUN benchmark resources

Published benchmarking results, analysis scripts, and sequencing resources. GEO GSE282809.

Explore GEO data

Epigenome Lab code repositories

Public research software and analysis code, including the CUT&RUN peak-calling benchmark.

Browse GitHub