Periodic Reporting for period 1 - UBioRec (Development and Testing of a Reference Computational Platform for Understanding BiomolecularRecognition)
Reporting period: 2018-06-01 to 2020-05-31
The binding of ligands to pharmaceutical targets is a dynamical event and molecular simulations are uniquely powerful in their ability to track the behavior of ligand-target interactions at atomic resolution across different time scales. The promise of the approach has been demonstrated by the successful MD-based design of new drugs such as HIV integrase inhibitors. Nowadays, thanks to improved algorithms and hardware MD simulations are used to address the many open questions about the details of molecular recognition between protein-ligand complexes that affects the field of Drug Discovery, becoming an essential tool for the discovery of new medicines.
However, accurate molecular dynamics simulations are expensive and the right balance between accuracy and time to solution has to be found to increase their usefulness in drug discovery. The aim of our project was to further develop and test simulations-based models and algorithms to find an effective balance for drug discovery.
The Covid-19 pandemic has dramatically reaffirmed the importance of science in general and drug discovery in particular for the society.
The UBioRec project by providing effective computational algorithms to study drug binding mechanisms and accurately quantify the binding affinity of ligands to their targets addresses a fundamental aspect of rational drug discovery. The platform that has been designed, developed and made available during the project can be used by academic and industrial groups to accelerate the development of drug candidates. The advantage of the molecular dynamics based algorithms developed in UBioRec, with respect to static (docking based) ones that are typically used is that they take into account all the structural changes that occur during the binding process, and therefore are more accurate and lead to a better understanding of the binding mechanism.
This fundamental knowledge is crucial to efficiently develop potent and effective drug candidates.
The overall objectives of UBioRec project were the following:
• Test and combine different enhanced sampling algorithms with the most recently developed force fields and by means of QM/MM calculations. The achievement of this aim let us to comprehend the binding mechanism of the reference targets selected.
• Understand the thermodynamics and kinetics of the binding mechanisms.
• To make the methods and results obtained widely available to the scientific community through a platform. The data generated such as inputs, scripts and algorithms developed have been made publicly available in different repositories. Also, the benchmark data will be available through a platform where all the data will be available. The performance, maintenance and update of this platform will be carried out beyond to the finish of this project in order to become a much-needed reference for academic and industrial groups in the field.
However, accurate molecular dynamics simulations are expensive and the right balance between accuracy and time to solution has to be found to increase their usefulness in drug discovery. The aim of our project was to further develop and test simulations-based models and algorithms to find an effective balance for drug discovery.
The Covid-19 pandemic has dramatically reaffirmed the importance of science in general and drug discovery in particular for the society.
The UBioRec project by providing effective computational algorithms to study drug binding mechanisms and accurately quantify the binding affinity of ligands to their targets addresses a fundamental aspect of rational drug discovery. The platform that has been designed, developed and made available during the project can be used by academic and industrial groups to accelerate the development of drug candidates. The advantage of the molecular dynamics based algorithms developed in UBioRec, with respect to static (docking based) ones that are typically used is that they take into account all the structural changes that occur during the binding process, and therefore are more accurate and lead to a better understanding of the binding mechanism.
This fundamental knowledge is crucial to efficiently develop potent and effective drug candidates.
The overall objectives of UBioRec project were the following:
• Test and combine different enhanced sampling algorithms with the most recently developed force fields and by means of QM/MM calculations. The achievement of this aim let us to comprehend the binding mechanism of the reference targets selected.
• Understand the thermodynamics and kinetics of the binding mechanisms.
• To make the methods and results obtained widely available to the scientific community through a platform. The data generated such as inputs, scripts and algorithms developed have been made publicly available in different repositories. Also, the benchmark data will be available through a platform where all the data will be available. The performance, maintenance and update of this platform will be carried out beyond to the finish of this project in order to become a much-needed reference for academic and industrial groups in the field.
The work performed within the UBioRec project was divided into different work packages.In the first step, the objectives were to test and improve force fields parameters. Special attention has been considered for the quality of the atomic charges and torsional potentials, as they are expected to be crucial for the energetics of ligand binding/unbinding to/from the binding sites of targets selected. The use of a99SB-disp force field for the targets was corroborated as the best option in comparison with different force fields realized along the first semester of the project.
Then, the main work package involve the testing and developing of different enhanced-sampling algorithms for the calculation of binding free energies, with a good balance between accuracy, computational cost and speed. The main algorithms used were based on funnel-shaped restraint metadynamics. The advantage of using funnel-shaped restraints is the increase it provides to the speed of convergence of the simulations, and favouring re-crossings between bound and unbound states. Additionally, the general CVs that define the funnel restraints avoid extra simulations, reducing the number of computational resources needed for more computational demanding CVs. The combination of fun-metaD with SWISH (Sampling Water Interfaces through Scaled Hamiltonians), a Hamiltonian replica-exchange based method recently developed by Gervasio’s group was also used.During the UBioRec project, a novel application of the method was applied helping to understand the hydration/dehydration process during the dynamics itself in an unprecedented manner. An additional crucial idea developed in the UBioRec project is the definition of general convergence criteria, which help us to objectively evaluate the convergence of enhanced sampling simulations, independently of the experimental binding affinities available. The criteria established are general and have been applied to all the methods mentioned above. From these results, I published two papers, one as a first author and one as corresponding author.
Additional steps were i) the combination of computational approaches at different scales, and ii) obtaining the experimental data by employing biophysical techniques. For the first point, the results from the previous ES simulations were used to get the most relevant clusters for the QM/MM re-weighting calculations. For the latest point, the expression and purification of WT and the mutated targets were successfully obtained. SPR experiments were run providing information about binding kinetics and affinity. These results were needed to complement and validate the theoretical results obtained during the developing of the project. Finally, for the creation of reference data and platform all my efforts were concentrated on the compilation of the data obtained during the UBioRec project, together with the scripts and inputs used during the simulation process. All this information was structured in a GitLab repository in order to organize and develop the platform. All the files required to reproduce the simulations have been uploaded to the Plumed-Nest repository. The source code for the new CVs developed, are now part of the development branch of the Plumed plugin and will be made available in the next releases for the whole community.
Then, the main work package involve the testing and developing of different enhanced-sampling algorithms for the calculation of binding free energies, with a good balance between accuracy, computational cost and speed. The main algorithms used were based on funnel-shaped restraint metadynamics. The advantage of using funnel-shaped restraints is the increase it provides to the speed of convergence of the simulations, and favouring re-crossings between bound and unbound states. Additionally, the general CVs that define the funnel restraints avoid extra simulations, reducing the number of computational resources needed for more computational demanding CVs. The combination of fun-metaD with SWISH (Sampling Water Interfaces through Scaled Hamiltonians), a Hamiltonian replica-exchange based method recently developed by Gervasio’s group was also used.During the UBioRec project, a novel application of the method was applied helping to understand the hydration/dehydration process during the dynamics itself in an unprecedented manner. An additional crucial idea developed in the UBioRec project is the definition of general convergence criteria, which help us to objectively evaluate the convergence of enhanced sampling simulations, independently of the experimental binding affinities available. The criteria established are general and have been applied to all the methods mentioned above. From these results, I published two papers, one as a first author and one as corresponding author.
Additional steps were i) the combination of computational approaches at different scales, and ii) obtaining the experimental data by employing biophysical techniques. For the first point, the results from the previous ES simulations were used to get the most relevant clusters for the QM/MM re-weighting calculations. For the latest point, the expression and purification of WT and the mutated targets were successfully obtained. SPR experiments were run providing information about binding kinetics and affinity. These results were needed to complement and validate the theoretical results obtained during the developing of the project. Finally, for the creation of reference data and platform all my efforts were concentrated on the compilation of the data obtained during the UBioRec project, together with the scripts and inputs used during the simulation process. All this information was structured in a GitLab repository in order to organize and develop the platform. All the files required to reproduce the simulations have been uploaded to the Plumed-Nest repository. The source code for the new CVs developed, are now part of the development branch of the Plumed plugin and will be made available in the next releases for the whole community.
The progress made by the UBioRec project involves the release of novel algorithms and methods, establishing the areas in which each of these methods excels and provided guidelines on their application and the most appropriate settings. One of the most important outcomes represents the advances made in evaluating real complex systems, being able to tackle more flexible and complex targets and larger ligands. Regarding the socio-economic impact, the proposed project is built upon the recently renewed interest in computational drug discovery (e.g. see the $120M computational drug design deal with Schrödinger made by Sanofi or the $1.2 billion investment in Nimbus Therapeutics by Gilead), representing very significant progress in the field. The UBioRec project contributes to a faster and more efficient process of drug discovery, which could also help to reduce costs within the industry. Moreover, the society is now able to understand why it is important to improve the mechanisms by which a drug needs to speed up its own production.