Overview / Computational biology

Computational biology

I analyse single-cell, spatial and TCR sequencing data myself, from raw reads on the cluster through to published figures. Most of these datasets come from experiments I designed or ran.

public repositories
9
papers with code
5
languages
R, Python
compute
HPC

Single-cell and multi-omics analysis

Data types
scRNA-seq, CITE-seq, scATAC-seq and scTCR-seq; spatial transcriptomics (Visium) and high-multiplex imaging (CODEX/PhenoCycler); bulk RNA-seq and TCR-seq; exome and genome sequencing.
Processing and QC
CellRanger, STARsolo, Salmon/alevin-fry, kallisto-bustools; ambient RNA, doublet and empty-droplet removal (SoupX, CellBender, scDblFinder); hashtag and genotype demultiplexing.
Integration and annotation
Harmony, Seurat CCA/RPCA and weighted nearest neighbour; reference mapping and label transfer (Azimuth, CellTypist, SingleR, ProjecTILs).
Differential testing
Pseudobulk differential expression (DESeq2, edgeR, limma-voom); differential abundance (Milo); pathway and enrichment analysis (GSEA, decoupleR, PROGENy).
Trajectories and lineage
RNA velocity (scVelo, CellRank); pseudotime (Monocle, Slingshot); lineage tracing with TCR clones and mitochondrial DNA variants (mgatk, ReDeeM).
Gene regulation
Regulatory networks (SCENIC, DoRothEA); co-expression modules (WGCNA); chromatin accessibility with ArchR and Signac.
Immune repertoires
Clonotype calling (CellRanger VDJ, TRUST4, MiXCR), diversity and clonal tracking, CDR3 properties, specificity grouping (GLIPH, TCRdist), TCR embeddings (TCR-BERT).
Communication and space
Cell–cell communication (CellChat, NicheNet, LIANA); spatial neighbourhood and proximity analysis.
Other
Copy-number inference (inferCNV, CopyKAT); scVI and single-cell foundation models (scGPT, Geneformer).

Statistics, programming and compute

Languages
R (primary: tidyverse, ggplot2, Quarto); Python and shell scripting developing
Statistics
Linear and mixed-effects models, GLMs and GAMs; survival analysis; compositional data; experimental design, power and multiple-testing correction; permutation, bootstrap and non-parametric tests.
HPC and environments
SLURM (job arrays, dependency chains, resource estimation); Apptainer/Singularity and Docker; renv and uv environments; Git and GitHub.
Pipelines
Running nf-core and Snakemake pipelines; writing my own developing
Public data
GEO, ENA/SRA, ArrayExpress, EGA, CELLxGENE, TCGA and GTEx; FAIR deposition to Zenodo.
Software
R package development with tests, CI and documentation (r-utility-functions).
AI-assisted research
Agentic coding and analysis with Claude Code, including writing skills and running subagents.

Developing

Areas I am still building. Public work is linked where it exists.

Protein structure
Structure prediction (AlphaFold3, Boltz-2, Chai-1), homology search (Foldseek), pore analysis, de novo binder design (RFdiffusion, ProteinMPNN). See the NKG7 repositories below.
Genomic variants
Germline and somatic variant calling (GATK, Mutect2), copy number (CNVkit), annotation (gnomAD, ClinVar), HLA typing and neoantigen prediction.
Machine learning
Applied machine learning for outcome prediction; deep learning.

Code behind the papers

Analysis code I wrote for published papers, one repository per paper.

RepositoryPaperYear
Turiello_Ng_etal_EJI_2025Turiello, Ng et al. European Journal of Immunology2025
Corvino_etal_2024_EJICorvino et al. European Journal of Immunology2025
Giordano_Layer_Leonardelli_etal_2024_Nature_ComsGiordano, Layer, Leonardelli et al. Nature Communications2024
Hosni_Kilian_etal_Cancer_Research_2023Hosni, Kilian et al. Cancer Research2024
Braun_et_al_Immunity_2020Braun et al. Immunity2020

Tools and current projects

RepositoryWhat it does
r-utility-functionsRR package of reusable bioinformatics & single-cell helper functions (heatmaps, volcano plots, GO, UMAP optimisation, TCGA survival, colour schemes)
Thymic_NK_developmentPythonDo NK cells carry a thymic TCR footprint, and are CD56bright and CD56dim NK distinct lineages? Single-cell VDJ and somatic mtDNA lineage tracing across public datasets
nkg7-binder-designPythonDe novo mini-binder design pipeline (RFdiffusion, ProteinMPNN, Boltz-2, Chai-1) applied to NKG7 extracellular loops, with a sequence-complexity audit of the designs
nkg7-structural-analysisPythonIn-silico structural & evolutionary characterisation of NKG7 (cytotoxic-lymphocyte granule tetraspan): structure prediction + homology + pore analysis. Tests the ion-channel hypothesis — finds a self-associating claudin/Ca_V-γ scaffold, not a pore.

All public repositories: github.com/Eomesodermin.