Computational activity screening of enzymes (CAZE)
Main funder
Funder's project number: 342908
Funds granted by main funder (€)
- 244 130,00
Funding program
Project timetable
Project start date: 01/01/2022
Project end date: 31/12/2024
Summary
Enzymes – proteins that catalyze chemical reactions – are widely used for industrial purposes (food making, laundry, pharmaceutics, etc.). Developing new, more efficient enzymes helps to make services cheaper and industrial production eco-friendlier. In this project, I am going to develop a set of computational tools, that might ease the design of new enzymes. In contrast to currently used workflows, that frequently concentrate on enzyme amino acid sequences, I will focus on the structural properties, obtained in molecular dynamics (MD) or quantum mechanics/molecular mechanics (QM/MM) simulations.
MD simulations allow to track conformational changes of enzymes, by computing forces acting on every atom and integrating equations of motion. However, its main limitation is inability to model chemical reactions. QM/MM simulations combine quantum mechanical description of reactive cite with classical description of the rest of the system. They allow chemical reactivity, but the sampling is poor. To efficiently use structural methods for enzyme design, the sufficient sampling, available in MD, and chemical reactivity, available in QM/MM, should both be available.
During the project, I will extend classical MD simulations to allow chemical reactivity. That has been done for Lammps software by including ReaxFF package. I will develop a reactive package compatible with Gromacs software (www.gromacs.org). The package will be based on -dynamics constant pH protocol, available in Gromacs. Currently it allows only the electrostatic interactions to change and does not introduce computational cost, if compared to classical MD. For chemical reactions, I will extend it to allow Lennard-Jones and bonded interactions to change as well.
The advantage of such reactive protocol, is that it will be easy to use: no additional parameters should be provided, in contrast to ReaxFF, where the functional forms of interaction is different from those in standard MD. However, that will lead to limitations: (i) wrong chemical intermediates might be sampled; (ii) incorrect free energy landscape might be predicted. To overcome these limitations, I will develop a PATHFINDER tool, that will find the lowest-barrier pathway through intermediates using Gromacs-CP2K interface (https://bioexcel.eu/software/cp2k/) developed in the host group.
Finally, reactive simulations and PATHFINDER will be enveloped in two automatic workflows: CompAct to predict enzyme activity and CompARe for the computational screening of enzyme mutants, to find the most efficient ones. These workflows will be tested in collaboration with Novozymes company. If successful, the developed approach will allow the efficient and cheap, in terms of human resources needed, workflow for enzyme design. The developed tools will not only be a huge leap forward for MD simulations, but also might be appropriated by biotechnological and pharmaceutical companies.
MD simulations allow to track conformational changes of enzymes, by computing forces acting on every atom and integrating equations of motion. However, its main limitation is inability to model chemical reactions. QM/MM simulations combine quantum mechanical description of reactive cite with classical description of the rest of the system. They allow chemical reactivity, but the sampling is poor. To efficiently use structural methods for enzyme design, the sufficient sampling, available in MD, and chemical reactivity, available in QM/MM, should both be available.
During the project, I will extend classical MD simulations to allow chemical reactivity. That has been done for Lammps software by including ReaxFF package. I will develop a reactive package compatible with Gromacs software (www.gromacs.org). The package will be based on -dynamics constant pH protocol, available in Gromacs. Currently it allows only the electrostatic interactions to change and does not introduce computational cost, if compared to classical MD. For chemical reactions, I will extend it to allow Lennard-Jones and bonded interactions to change as well.
The advantage of such reactive protocol, is that it will be easy to use: no additional parameters should be provided, in contrast to ReaxFF, where the functional forms of interaction is different from those in standard MD. However, that will lead to limitations: (i) wrong chemical intermediates might be sampled; (ii) incorrect free energy landscape might be predicted. To overcome these limitations, I will develop a PATHFINDER tool, that will find the lowest-barrier pathway through intermediates using Gromacs-CP2K interface (https://bioexcel.eu/software/cp2k/) developed in the host group.
Finally, reactive simulations and PATHFINDER will be enveloped in two automatic workflows: CompAct to predict enzyme activity and CompARe for the computational screening of enzyme mutants, to find the most efficient ones. These workflows will be tested in collaboration with Novozymes company. If successful, the developed approach will allow the efficient and cheap, in terms of human resources needed, workflow for enzyme design. The developed tools will not only be a huge leap forward for MD simulations, but also might be appropriated by biotechnological and pharmaceutical companies.
Principal Investigator
Other persons related to this project (JYU)
Primary responsible unit
Follow-up groups
Related publications and other outputs
- Approaching Optimal pH Enzyme Prediction with Large Language Models (2024) Zaretckii, Mark; et al.; A1; OA
- gmXtal : Cooking Crystals with GROMACS (2024) Buslaev, Pavel; et al.; A1; OA
- Overlay databank unlocks data-driven analyses of biomolecules for all (2024) Kiirikki, Anne M.; et al.; A1; OA
- phbuilder : A Tool for Efficiently Setting up Constant pH Molecular Dynamics Simulations in GROMACS (2024) Jansen, Anton; et al.; A1; OA
- Best Practices in Constant pH MD Simulations : Accuracy and Sampling (2022) Buslaev, Pavel; et al.; A1; OA
