Chlorite Dismutase Oxygen Release Rescues Growth of Bacillus subtilis Under Oxygen Limitation: A Genome-Scale Modelling Study of a Mars-Relevant Perchlorate Pathway
FEMS Microbiology Letters.
My research sits at the intersection of computational oncology, spatial genomics, and multiomics, and asks why the same molecular intervention does not produce the same outcome in every patient. I trained as a physician first, and that shapes most of what I choose to work on: the computational questions I care about usually started as a clinical one, somewhere on a ward in Osogbo.
Research Focus
The same molecular intervention produces divergent outcomes in different patients, and the divergence is signal, not noise. My work maps that signal using spatial genomics and single-cell multiomics, and builds the computational infrastructure to act on it.
Three threads converge on the problem. DEEPTNBC applies a hybrid CNN-BiLSTM architecture to model PARP1/BRCA1 co-targeting in triple-negative breast cancer. ROOTS screens natural products against ovarian cancer targets through structural and pharmacokinetic filtering. A structural bioinformatics pipeline I built caught an accession error in the Regolyx dataset before it propagated downstream, the kind of failure that looks minor until it corrupts every result built on it. At MIT Media Lab's How to Grow Almost Anything programme, ÌṢỌ-Sentinel EcN engineered a whole-cell sentinel biosensor architecture and placed in the upper cohort tier.
Biological systems are heterogeneous at every scale, genome to cell to population, and the tools built to study them must be heterogeneity-aware from the ground up.
Nigeria carries a disproportionate share of the global cancer burden against a fraction of the computational infrastructure wealthier systems assume as baseline. Precision oncology has largely been built on data that does not reflect African populations. Closing that gap requires people trained at the frontier who build the bridge back. An MBBS and clinical experience in Osogbo keep the work concrete.
Manuscripts under review or in preparation
FEMS Microbiology Letters.
With A.T. Ojo. Medical Microbiology and Immunology, Springer Nature.
Current Microbiology, Springer Nature.
A hybrid CNN-BiLSTM pipeline for ranking drug candidates that co-target PARP1 and BRCA1, validated through docking and molecular dynamics.
Omics-driven virtual screening of natural products against multi-target ovarian cancer, combining docking, MD simulation, and MM-PBSA binding free energy analysis. ASBCB Omics Codeathon.
Automated first-pass quality control, clustering, and cell-type annotation reporting for single-cell and spatial transcriptomics data, built for wet-lab researchers and bioinformatics cores without in-house support.
Confidence-aware binding-pocket triage and molecular dynamics setup for predicted protein structures, pairing druggability scoring with the prediction's own per-residue confidence.
Peer-reviewed articles
With C.U. Emele.
Preprints & reports
With M. Denkiewicz, R.M. Ahmad, R. Gupta, and others. BioHackrXiv.
With T. Sisodiya, N. Rihan, and A. Udawat. Apart Research.
Recent projects
MIT Media Lab, How to Grow Almost Anything. Built a model-first framework treating cell state as a quantifiable, modelable property; designed biosensor (TtrS-TtrR), effector (MccH47), and biocontainment (ΔdapA) computationally; produced fitness-efficacy tradeoff maps via ODE modelling (Tellurium), SALib/PRCC, and Moran process. Individual final project ranked in the upper cohort tier.
EBI Metagenomics x EVBC nf-core Hackathon. Built a five-stage Nextflow DSL2 subworkflow moving metavirome characterisation from taxonomy to structural phenotype, integrating ESMFold, Foldseek (351,242 viral structures), ECOD classification, and ProteinCartography UMAP visualisation. Merged via peer-reviewed pull request following full nf-core community standards.
nf-core x VirJenDB x EVBC Hackathon. Contributed a specification-compliant Nextflow DSL2 module wrapping the GRIMER contamination detection engine for metagenomic QC, passing all 51 nf-core lint tests and pre-commit hooks prior to PR submission.
Baylor College of Medicine 7th Pan-Structural Variation Hackathon. Automated transposable-element annotation pipeline for structural variants from long-read sequencing data, combining RepeatMasker analysis with genomic intersection approaches (BEDTools, Snakemake).
Independent synthetic biology project. Genome-scale flux balance analysis showing chlorite dismutase oxygen release rescues Bacillus subtilis growth under oxygen limitation, in a Mars-relevant perchlorate biodegradation pathway. Engineered gene circuit design and metabolic pathway modelling for operation under low temperature and high radiation, using AutoDock, AMDock, and SnapGene.
PhD applications
Applying for Fall 2027 entry, positioning around computational oncology, spatial and single-cell multiomics, and the tumour immune microenvironment.