Katona group

Structural biology, protein dynamics and artificial intelligence

Research interests

During the last decade our research has converged on three connected themes: structural biology of dynamic proteins, data-driven analysis of biophysical experiments, and the molecular logic of protein interactions in chromatin and disease. The common aim is to connect molecular structure, dynamics and interaction energetics to biological function.

1. Structural biology of protein dynamics and reactions

We use crystallographic and X-ray scattering approaches to observe proteins away from static end states. A central theme has been the development and use of chemical, physical and light-triggered experiments for time-resolved structural biology. Recent work includes resampling-based interpretation of time-resolved serial X-ray crystallography, high-resolution macromolecular crystallography at FemtoMAX, structural changes in photosynthetic reaction centres, active-site interactions that shape enzyme temperature dependence, and long-range or terahertz-coupled dynamics in protein crystals.

Selected references

  • Vallejos, A., Katona, G. & Neutze, R. (2024) Appraising protein conformational changes by resampling time-resolved serial X-ray crystallography data. Structural Dynamics, 11, 044302.
  • Dods, R. et al. (2020) Ultrafast structural changes within a photosynthetic reaction centre. Nature, 589, 310-314.
  • Jensen, M. et al. (2021) High-resolution macromolecular crystallography at the FemtoMAX beamline with time-over-threshold photon detection. Journal of Synchrotron Radiation, 28, 64-70.
  • Winter, S. D. et al. (2021) Chemical mapping exposes the importance of active site interactions in governing the temperature dependence of enzyme turnover. ACS Catalysis.

2. Protein interactions, survivin, chromatin and immune disease

We study protein-protein interactions both as chemical problems and as functional networks. A major line concerns survivin/BIRC5, its peptide binding preferences, chromatin association and effects in CD4+ cells and rheumatoid arthritis. Recent work links survivin to PRC2 inhibition, BRG1/SWI chromatin remodelling, DNA damage response, glycolytic regulation and autoimmune phenotypes.

Selected references

  • Chandrasekaran, V. et al. (2024) Bivalent chromatin accommodates survivin and BRG1/SWI complex to activate DNA damage response in CD4+ cells. Cell Communication and Signaling, 22, 440.
  • Anindya, A. L. et al. (2024) Deciphering peptide-protein interactions via composition-based prediction: a case study with survivin/BIRC5. Machine Learning: Science and Technology, 5, 025081.
  • Jensen, M. et al. (2023) Survivin prevents the polycomb repressor complex 2 from methylating histone 3 lysine 27. iScience, 26, 106976.
  • Erlandsson, M. C. et al. (2022) Survivin promotes a glycolytic switch in CD4+ T cells by suppressing the transcription of PFKFB3 in rheumatoid arthritis. iScience, 25, 105526.
  • Chandrasekaran, V. et al. (2022) Cohesin-mediated chromatin interactions and autoimmunity. Frontiers in Immunology, 13, 840002.

3. Bayesian, machine-learning and mean-field models for biophysical data

We develop interpretable models rather than using prediction alone. This includes Bayesian analysis of MST, BLI, NMR and crystallographic experiments, machine-learning assessment of atomic displacement parameters and phasing, composition-based prediction of protein interactions, and mean-field descriptions of biomolecular association. The current goal is to connect these effective parameters to collective dynamics in macromolecular systems.

Selected references

  • Anindya, A. L. et al. (2024) Deciphering peptide-protein interactions via composition-based prediction: a case study with survivin/BIRC5. Machine Learning: Science and Technology, 5, 025081.
  • Anindya, A. L. et al. (2022) Bayesian progress curve analysis of MicroScale thermophoresis data. Digital Discovery, 1, 325-332.
  • Gagnér, V. A., Jensen, M. & Katona, G. (2021) Estimating the probability of coincidental similarity between atomic displacement parameters with machine learning. Machine Learning: Science and Technology, 2, 035033.
  • Garcia-Bonete, M.-J. & Katona, G. (2019) Bayesian machine learning improves single-wavelength anomalous diffraction phasing. Acta Crystallographica Section A, 75, 851-860.
  • Katona, G., Garcia-Bonete, M.-J. & Lundholm, I. V. (2016) Estimating the difference between structure-factor amplitudes using multivariate Bayesian inference. Acta Crystallographica Section A, 72, 406-411.