I am Haoting Zhang, a Postdoctoral Researcher at the National Heart and Lung Institute (NHLI), Imperial College London, mentored by Dennis Wang. I am working on developing machine learning methods for drug discovery, with an emphasis on improving their translational impact in clinical practice.

Previously, I completed my PhD as part of the 2020 cohort of the HDR UK–Turing Wellcome PhD Programme in Health Data Science, based at ML@CL in the Department of Computer Science and Technology, University of Cambridge. I was supervised by Carl Henrik Ek, Marta Milo, and Magnus Rattray. Prior to that, I studied Mathematics and Statistics as an undergraduate at St Hilda's College, University of Oxford, followed by a Master’s degree in Machine Learning at UCL.

For my PhD thesis, I am working on the inference and prediction of drug combination effects (in terms of synergy, potency and efficacy) from a Bayesian perspective. For the inference phase, we estimated the synergy of the drug combinations using a Bayesian framework so that parameter uncertainty can be estimated and existing knowledge can be inserted through the priors. Another motivation is that, through the uncertainty derived from the model, actionable decisions can be made to guide the subsequent experiments in the pre-clinical drug discovery pipeline. To this end, we developed SynBa, a flexible Bayesian approach to estimate the uncertainty of the synergistic efficacy and potency of drug combinations:

• Zhang, H., Ek, C. H., Rattray, M., & Milo, M. (2023). SynBa: Improved estimation of drug combination synergies with uncertainty quantification. Bioinformatics, 39(Supplement_1), i121-i130.

I am delighted to have presented this work at the ISMB/ECCB 2023 Conference. The code for SynBa contains notebooks that illustrate how users can upload their own monotherapy/combination dose-response data, fit SynBa to their data and visualise the inference output.

For the prediction phase, through collaboration, we developed DeepSynBa, a deep network that predicts the complete dose-response matrix of drug pairs instead of relying on an aggregated synergy score, motivated by our observation that a single synergy score is inadequate for capturing the full landscape of drug combination effects:

• Kuru, H.I.*, Zhang, H.*, Rattray, M., Ek, C.H., Cicek, A.E., Tastan, O. and Milo, M., 2025. DeepSynBa: Actionable Drug Combination Prediction with Complete Dose-Response Profiles. bioRxiv, pp.2025-01. (*: Equal Contribution)

Prior to my PhD thesis and during the first year of my PhD in the HDR UK programme, I worked jointly on multi-partition clustering of high-dimensional data (in particular images) through variational autoencoders with a hierarchy of latent variables, each with a Mixture-of-Gaussians prior. This work has been accepted at NeurIPS 2021:

• Falck, F.*, Zhang, H.*, Willetts, M., Nicholson, G., Yau, C., & Holmes, C. C. (2021). Multi-facet clustering variational autoencoders. Advances in Neural Information Processing Systems, 34, 8676-8690. (*: Equal Contribution)

Please see About Me for more details about myself :)