Publications

Bansal, J.K. and Nichols, R.A. (2025). Can genomic analysis actually estimate past population size? Trends in Genetics, 41(7):559–567. DOI: 10.1016/j.tig.2025.04.005 (Open access)

Research outputs

Research Assistant — Simulation & Analytical Modelling

Imperial College London · March 2026 – February 2027

Developed a large-scale stochastic simulation framework in Python to generate high-quality synthetic datasets for training machine learning surrogate models — combining classical probabilistic reasoning with data-driven approaches to improve scalability.

• Ran sensitivity and scenario analyses across 10,000+ parameter combinations on HPC infrastructure, validating synthetic outputs against empirical data to characterise model behaviour
• Built automated, modular Python pipelines adopted by collaborating teams, with careful attention to dataset quality, distributional coverage, and reproducibility
• Communicated model outputs, assumptions, and limitations clearly to both technical and non-technical audiences as requirements and biological assumptions evolved mid-project

PhD Researcher — Probabilistic & Statistical Modelling

Queen Mary University of London · September 2022 – September 2026

Doctoral research focused on extracting reliable signal from complex, noisy biological datasets — selecting between ODEs, stochastic simulation, Bayesian inference, hidden Markov models, and transition matrix approaches based on what the data and problem structure demanded, rather than convention.

Projects:

• Can genomic analysis actually estimate past population size?
• Estimates of IBD vary with recombination and mutation rate
• Using genetic data to learn about malaria transmission dynamics
• The distribution of autozygous segments in first cousins

Highlights:

• Designed controlled computational experiments to probe where and why methods break down: comparing simulation outputs against mathematical and statistical benchmarks to validate behaviour, characterise failure modes, and understand model limitations
• Collaborated with and supervised a PhD student on empirical datasets, testing whether theoretical models held up against real-world patterns across teams with varied technical backgrounds
• Peer-reviewed publication (Bansal & Nichols, Trends in Genetics, 2025; 13 citations): identified conditions under which a widely-used inference tool produces misleading signals, and developed strategies to distinguish genuine effects from artefacts — with a fully documented, reproducible GitHub repository
• Invited talk at LSHTM; Best Talk at the QMUL Third Year Symposium

For collaboration or questions about my work, get in touch.

---

Janeesh Kaur Bansal