Natasha Latysheva, PhD
I am a researcher in AI for biology at Google DeepMind, working on genomics, DNA sequence modelling, and variant effect prediction. My work focuses on building and analysing machine learning systems to understand and predict regulatory biology at scale.
​
Alongside research, I focus on teaching, outreach, and scientific communication, including co-authoring the O’Reilly book "Deep Learning for Biology" and delivering workshops internationally.
Research
AlphaGenome
​
Co-first author on DeepMind’s flagship genomics model, AlphaGenome, which predicts the regulatory effects of DNA sequence variation across multiple modalities, including gene expression and splicing.
​
The model enables state-of-the-art variant effect prediction zero-shot across modalities, providing a unified framework for modelling regulatory biology at scale.
​​​​​​
​Press​
​
Spokesperson for AlphaGenome, contributing to media coverage and public communication.
Research Roundtable​
Featured in an author roundtable with the AlphaGenome team discussing model design, applications, and future directions. [Watch on YouTube]
Recent Invited Talks​
​
-
Keynote, Annual Computational Biology Symposium (2026)
Cambridge, UK · [Link]
​​
-
Imperial College London, Department of Infectious Disease (2026)
London, UK · [Link]
​​
-
European Society of Human Genetics (ESHG) (2026)
Gothenburg, Sweden · [Link]
​​
-
Sir Nick Clegg Roundtable on AI for cancer research (2026)
London, UK · [Link]
​​
-
Keynote, Benchling Digital Science & Innovation Forum (2026)
London, UK · [Link]​
​
Full list of talks → please contact me.
Selected Research Publications​
-
Avsec, Z., Latysheva, N., …, Kohli, P. (2026)
Advancing regulatory variant effect prediction with AlphaGenome.
Nature [Link]
​​
-
Saab, K., …, Latysheva, N., …, Natarajan, V. (2024)
Capabilities of Gemini models in medicine.
arXiv [Link]
​​
-
Latysheva, N. S., …, Babu, M. M. (2016)
Molecular principles by which gene fusions affect protein interaction networks in cancer.
Molecular Cell [Link]
​​
-
Chavali, S., …, Latysheva, N. S., …, Babu, M. M. (2017)
Constraints and consequences of the emergence of amino acid repeats in eukaryotic proteins.
Nature Structural & Molecular Biology [Link]
​​
-
Latysheva, N. S., …, Babu, M. M. (2019)
Molecular signatures of fusion proteins in cancer.
ACS Pharmacology & Translational Science [Link]
​​​
Full publication list → [Google Scholar]
Teaching
Book
Deep Learning for Biology (O’Reilly), co-authored with Charles Ravarani, is a practical guide to applying modern machine learning methods to biological problems.
​
The book bridges core ML architectures - including CNNs, transformers, graph neural networks, and variational autoencoders - with real applications such as protein function prediction, regulatory genomics, cancer imaging, and drug–drug interaction modelling.
​
It includes hands-on JAX/Flax implementations, along with guidance on model interpretability and rigorous experimental design, with the aim of making modern machine learning accessible to biology researchers and students.
Buy the book at:
​
Check out the [Github repo] for the accompanying Colab notebooks for each chapter.
Selected Workshops and Practicals
I design and deliver workshops on machine learning for biology, focusing on making modern methods accessible to diverse research communities. I teach in international programmes including Deep Learning Indaba (Africa), EEML (Eastern Europe), and MenaML (Middle East and North Africa), with a particular emphasis on supporting early-career scientists and students from underrepresented groups.
​
-
MenaML | AI for Biology | 2026 | KAUST, Saudi Arabia | [Link]
-
University of Exeter | Unmasking the Genome: Integrating WGS, AI and Functional Genomics | Exeter, UK | [Link]
-
EEML | Montenegrin Machine Learning Workshop | AI for Biology | 2025 | Podgorica, Montenegro | [Link]
- MenaML | AI for Genomics | 2025 | Doha, Qatar | [Watch on YouTube]
-
Deep Learning Indaba | AI for Biology Practical Lead | 2024 | Dakar, Senegal | [Link]
-
Deep Learning Indaba | AI for Biology Practical Lead | 2023 | Accra, Ghana | [Link]
-
Deep Learning Indaba | TA for various ML practicals | 2022 | Tunis, Tunisia
-
Google DeepMind | JAX Gym Internal Course | 2021-2023 | London, UK
-
Code First: Girls | Instructor for 8-week Python evening course | 2016 | Cambridge, UK
-
Cambridge Coding Academy | Machine Learning Workshops Instructor | 2016 | Cambridge, UK​
​​​
Full list of workshops → please contact me.
​​
About me
Experience
2020-Present
Senior Research Engineer
Google DeepMind
London, UK
Machine learning for genomics and molecular biology, focused on developing biological foundation models for DNA function and predicting the effects of genetic mutations. Applications involve improving our understanding of the genetic basis of human disease.
2018-2020
Machine Learning Engineer
Welocalize
London, UK / Dublin, Ireland
Launched the company’s in-house machine learning efforts for machine translation and other natural language problems. Led ML projects for several large clients. Ran tutorials and workshops on AI and ML.
2017
Data Scientist
Jagex Games Studio
Cambridge, UK
Data science and machine learning on large-scale video game data. Developed techniques for player clustering, bot detection, and sentiment analysis.
Academic Background
2013-2017
PhD Computational Biology
Trinity College
University of Cambridge
Doctoral thesis titled “Computational characterisation of gene fusion mutations in cancer”. Analysis and machine learning on cancer genetics data. Advised by Madan Babu at the Laboratory of Molecular Biology.
2009-2013
BSc Biochemistry (Honours)
University of St Andrews
First Class. Ranked top of class all four years. Additional classes in statistics and mathematics to supplement biology education. Awarded Principal's Medal for 'outstanding achievements'.