Computational and Machine Learning Scientist in Genomics
MRC Weatherall Institute of Molecular Medicine (WIMM)
University of Oxford

🔬 Milne Group – Epigenetics & Gene Regulation in Leukaemia

📍 Oxford, UK | 🧪 Computational & Experimental Epigenomics | 💡 Enhancer Biology in Cancer

I investigate enhancer dynamics in cancer, with a particular focus on MLL-AF4 Acute Lymphoblastic Leukaemia (ALL). My work sits at the interface of experimental epigenomics and computational biology, combining multi-modal high-throughput datasets generated from patient-derived samples and disease models to understand chromatin-driven transcriptional dysregulation in leukaemia.

To address these questions, I generate and integrate diverse high-throughput datasets using:

• Epigenomic profiling: ChIP-seq, low-input ChIPmentation/TOPmentation, ATAC-seq, CUT&Tag, CUT&RUN
• Transcriptomics: Bulk RNA-seq, TT-seq, POINT-seq
• Chromatin conformation: NG Capture-C, Tiled-C, Micro-Capture-C
• Epigenetic modifications: DNA methylation profiling - TAPS/Bisulphite based methods
• Single-cell multiomics: 10x scMultiome, scCUT&TAG

These data are analysed using custom workflows, statistical modelling, and machine learning. Examples include a CatBoost classifier with SHAP feature importance to identify features of patient-specific enhancers using ChIP-seq datasets, and a custom PyTorch autoencoder trained on patient ALL blast H3K27ac profiles to detect sample-specific regulatory elements at scale.

I build open-source tools designed for performance, reproducibility, and ease of use in large-scale genomics. All projects ship with Docker and Apptainer containers; publication-ready tools include full CI/CD workflows (automated testing, build, and release).

• 🔄 SeqNado — docs: Modular Snakemake pipelines for processing ChIP-seq, ATAC-seq, RNA-seq, CUT&Tag, WGS, CRISPR screens, and Micro-Capture-C (MCC) data (PyPI, Conda)
• 🧬 CapCruncher — docs: Toolkit for analysing NG Capture-C, Tri-C, and Tiled-C chromatin interaction data (PyPI, Conda)
• 📦 QuantNado: High-performance genomic signal quantification platform. Per-bp coverage stored in Zarr v3 + Dask for large-scale multi-omics access (BAM, methylation, variants). Peak calling via 5 methods including a custom 1D ResNet and U-Net (PyTorch, semi-supervised learning, Optuna hyperparameter optimisation). (PyPI)
• 🔬 ScNado: Rust + Python + Snakemake pipeline for single-cell CUT&Tag and RNA-seq, integrating Scanpy, SnapATAC2, and Muon for multimodal single-cell analysis
• 🤖 Greyhound: Fine-tuning framework for genomics foundation models (Borzoi, Enformer). LoRA/LoCon parameter-efficient fine-tuning, custom chromatin prediction heads, HuggingFace transformers-compatible. PyTorch Lightning + W&B + Hydra.
• ⚡ BamNado: A fast, Rust-based utility for efficient BAM file manipulation (crates.io, PyPI)
• 📊 PlotNado: Command-line genomic visualisation in a genome browser-like style (PyPI)
• 🌐 TrackNado: A Python package for efficient generation of UCSC Genome Browser hubs for rapid distribution and visualisation of genomic datasets (PyPI, Conda)

1. Smith AL et al. (2025) — Enhancer heterogeneity in acute lymphoblastic leukemia drives differential gene expression in patients — Blood
2. Lau I-J et al. (2026) — MYB activity drives emergent enhancer activation and enhancer-promoter interactions in acute lymphoblastic leukemia — Blood
3. Crump NT†, Smith AL† et al. (2023) — MLL-AF4 cooperates with PAF1 and FACT to drive high-density enhancer interactions in leukemia — Nature Communications
4. Downes DJ†, Smith AL† et al. (2022) — Capture-C: a modular and flexible approach for high-resolution chromosome conformation capture — Nature Protocols
5. Godfrey L et al. (2019) — DOT1L inhibition reveals a distinct subset of enhancers dependent on H3K79 methylation — Nature Communications
6. Gao Z†, Smith AL† et al. (2023) — Temporal analyses reveal a pivotal role for sense and antisense enhancer RNAs in coordinate immunoglobulin lambda locus activation — Nucleic Acids Research

• Languages: Python, Rust, R, Bash, Snakemake
• ML Frameworks: PyTorch, PyTorch Lightning, TensorFlow/Keras, HuggingFace Transformers, CatBoost, SHAP, Optuna
• ML Practices: LoRA/LoCon fine-tuning, semi-supervised learning, experiment tracking (W&B), config management (Hydra)
• Single-cell: Scanpy, ArchR, SnapATAC2, Muon, Signac/Seurat
• Bioinformatics: DESeq2/pyDESeq2, MACS2/3, SEACR, deepTools, featureCounts, MultiQC, Ensembl VEP, GATK, bcftools
• Data Infrastructure: Dask, Zarr, Polars, pyranges
• Workflow & Infra: Snakemake, Conda, Docker, Apptainer, Git, GitHub Actions (CI/CD), HPC (SLURM), AWS
• Domains: Epigenomics, transcriptomics, chromatin architecture, multi-omics integration, single-cell epigenomics, foundation model fine-tuning, variant effect prediction

• 🔬 Milne Group @ Oxford
• 🧪 ORCID
• 📫 Email
• 💼 LinkedIn