Kuzma Khrabrov

I'm a Research Scientist in Machine Learning and Chemistry, focusing on developing new methods for molecular modeling and drug discovery. My work focuses on deep learning approaches for molecular property prediction and generative models for drug design.

I'm particularly interested in quantum chemistry and machine learning methods for accurate molecular energy predictions. I led the development of ∇2DFT, a universal quantum chemistry dataset and benchmark for neural network potentials, as well as pioneering work on using generative adversarial networks for de novo drug design through the druGAN project.

My current research explores ways to improve molecular conformer generation and energy prediction using machine learning. I work on developing novel architectures and training approaches that can better capture the underlying physics and chemistry of molecular systems while maintaining computational efficiency. I'm passionate about bridging the gap between deep learning and chemistry to accelerate drug discovery and materials science.

Publications

nablaDFT: Large-Scale Conformational Energy and Hamiltonian Prediction benchmark and dataset.

Kuzma Khrabrov, Ilya Shenbin, A. Ryabov, Artem Tsypin, Alexander Telepov, Anton M. Alekseev, Alexander Grishin, P. Strashnov, P. Zhilyaev, S. Nikolenko, Artur Kadurin

Physical Chemistry, Chemical Physics - PCCP 2022

ABS

∇2DFT: A Universal Quantum Chemistry Dataset of Drug-Like Molecules and a Benchmark for Neural Network Potentials

Kuzma Khrabrov, Anton Ber, Artem Tsypin, Konstantin Ushenin, Egor Rumiantsev, Alexander Telepov, Dmitry Protasov, Ilya Shenbin, Anton M. Alekseev, M. Shirokikh, Sergey I. Nikolenko, E. Tutubalina, Artur Kadurin

Neural Information Processing Systems 2024

ABSPDF

FREED++: Improving RL Agents for Fragment-Based Molecule Generation by Thorough Reproduction

Alexander Telepov, Artem Tsypin, Kuzma Khrabrov, Sergey Yakukhnov, Pavel Strashnov, Petr Zhilyaev, Egor Rumiantsev, Daniel Ezhov, Manvel Avetisian, Olga Popova, Artur Kadurin

Trans. Mach. Learn. Res. 2024

ABSHTMLPDF

Gradual Optimization Learning for Conformational Energy Minimization

Artem Tsypin, L. Ugadiarov, Kuzma Khrabrov, Manvel Avetisian, Alexander Telepov, Egor Rumiantsev, Alexey Skrynnik, Aleksandr Panov, Dmitry Vetrov, E. Tutubalina, Artur Kadurin

International Conference on Learning Representations 2023

ABSHTMLPDF

Doping position estimation for FeRh-based alloys

Egor Rumiantsev, Kuzma Khrabrov, Artem Tsypin, Nikita D Peresypkin, R. Gimaev, Vladimir I. Zverev, Roman Eremin, Artur Kadurin

Scientific Reports 2024

ABSPDF

druGAN: An Advanced Generative Adversarial Autoencoder Model for de Novo Generation of New Molecules with Desired Molecular Properties in Silico.

Artur Kadurin, S. Nikolenko, Kuzma Khrabrov, A. Aliper, A. Zhavoronkov

Molecular Pharmaceutics 2017

ABS

The cornucopia of meaningful leads: Applying deep adversarial autoencoders for new molecule development in oncology

Artur Kadurin, A. Aliper, A. Kazennov, Polina Mamoshina, Q. Vanhaelen, Kuzma Khrabrov, A. Zhavoronkov

OncoTarget 2016

ABSPDF

Chemical Language Models Have Problems with Chemistry: A Case Study on Molecule Captioning Task

Veronika Ganeeva, Kuzma Khrabrov, Artur Kadurin, Andrey V. Savchenko, E. Tutubalina

Tiny Papers @ ICLR 2024

The small cage in the Zoo of terminal Fano threefolds

Kuzma Khrabrov