Improving sampling of binding free energy differences between covalently bound ligands in alternate binding pockets using MT-REXEE
Anika J. Friedman, Michael R. Shirts · 2025
Read Improving sampling of binding free energy differences between covalently bound ligands in alternate binding pockets using MT-REXEE by Anika J. Friedman, Michael R. Shirts (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-06720.txt.
Introduction
Anika Friedman and co-authors improve a sampling method for drug design math in a paper from 2025. Chemists use free energy methods to guess how strongly a drug binds its target. Flexible systems are hard to sample, since high energy walls block the paths a simulation must cross.
Their test case is an enzyme called FabB and the carrier protein that feeds it. Growing chains on the carrier can sit in at least two pockets of the enzyme. The method, called MT-REXEE, swaps between states to sample both pockets well. It also handles ligands bound by a chemical bond. Better sampling means binding estimates that teams can trust more.
Catalog record
| CREATOR | Anika J. Friedman, Michael R. Shirts |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 61 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-06720
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Improving sampling of binding free energy differences between covalently bound ligands in alternate binding pockets using MT-REXEE Anika J. Friedman, Michael R. Shirts Abstract The primary limitation for the application of alchemical free energy methods to a wider variety of complex molecular systems in achieving reasonable sampling. Flexible binding complexes often have high free energy barriers, which require prohibitively long simulations to sample sufficiently to obtain reliable free energy estimates. An example of such a system is the FabB–acyl chain complex, which features acyl chain substrates from 4 to 16 carbons in length. Previous experimental evidence supports that there are also at least two accessible binding pockets for the acyl chain bound to FabB. Sampling of both pockets would require prohibitively long MD simulations or the creation of complicated and nonobvious collective variables. With the multiple topology replica exchange of expanded ensemble (MT-REXEE) enhanced sampling approach, we can obtain highly efficient sampling between both pockets by adaptively growing and shrinking the chains in the simulation ensemble, allowing each simulation to visit chain lengths where transitions between the pockets do occur. To enable this enhanced sampling for the chain growth problem, we demonstrate a new and improved conformational swapping selection algorithm for MT-REXEE, and validate this algorithm on a previously studied simultaneous calculation of multiple relative binding free energies. Using the new swapping approach gives significantly enhanced sampling even for this simpler problem, as demonstrated by faster convergence of free energy estimates. This case study demonstrates the utility of MT-REXEE and its open-source implementation for systems that feature high free energy barriers for a subset of ligands of interest, demonstrating a valuable addition to the existing stable of enhanced sampling methods. † † affiliation: Department of Chemical and Biological Engineering, University of Colorado Boulder, Boulder, CO 80309 † † affiliation: Department of Chemical and Biological Engineering, University of Colorado Boulder, Boulder, CO 80309 † † email: [email protected] 1 Introduction Alchemical free energy (FE) calculations have become standard in many computational molecular researchers’ workflows, but their application is primarily limited to systems with comparatively little conformational flexibility York 2023 ; Chodera et al. 2011 . Though the primary application of FE simulations has historically been small-molecule drug discovery Abel et al. 2017 ; Muegge and Hu 2023 , their utility extends far beyond such binding free energy calculations. However, these less standard applications often involve systems that exhibit significant conformational flexibility, either in the protein, the ligand, or both, requiring prohibitively long simulations to converge FE estimates. In many cases, systems with significant flexibility can remain trapped in metastable states for longer than is reasonable to simulate them. These limitations have prompted the discovery of numerous enhanced sampling methods, including our own multiple topology replica exchange of expanded ensemble Friedman et al. 2025 . Protein and enzyme engineering is an important, emerging application for FE calculations in large and flexible systems La Serra et al. 2022 ; Papadourakis et al. 2023 . Often, engineered protein systems involve multimeric complexes, which introduce flexible inter-protein interfaces. One such example is the enzymes within the fatty acid synthesis (FAS) cycle. This cycle is of interest for the industrial biosynthesis of a wide range of products, including bio-fuel precursors Singh et al. 2022 ; Pfleger and Takors 2023 . FAS involves enzymes that form either dimers or tetramers. The substrate for FAS enzymes is transported between each enzyme in the cycle by the acyl carrier protein (ACP) Buyachuihan et al. 2024 ; Chan and Vogel 2010 ; Grininger 2023 ; Meier and Burkart 2009 ; Walsh 2016 . Thus simulation of the ACP–enzyme complex involves multiple flexible binding interfaces. The covalent intermediate for FAS is both biologically relevant and a more straightforward state to quantify sampling enhancements than the ACP–enzyme complex. The covalent intermediate forms after the acyl chain substrate has been passed to the CYS residue on the respective FAS enzyme. The covalent intermediate state for FAS enzymes suffers from poor sampling in MD simulations due to the presence of a large, flexible substrate whose conformational flexibility is restricted by the covalent bonding to the protein and a relatively narrow binding pocket. These same sampling challenges also occur when applying FE simulations to covalent drug development De Cesco et al. 2017 ; Muegge and Hu 2023 ; Schaefer and Cheng 2023 ; Singh et al. 2011 as covalent drugs have a similar covalent intermediate. Covalent drug development is an emerging field given the increased potency and lifetime within the system. However, enhancing the sampling of AFE methods by capturing multiple binding and unbinding events as alchemical parameters change to minimize the configurational free energy barriers is not possible for covalent ligands using molecular mechanics simulations, as covalent bonds cannot be formed or broken. FE calculations with covalent ligands, both in the enzyme engineering and covalent drug cases, thus require alternate methods of achieving enhanced sampling to improve the accuracy and utility of computational methods. Existing enhanced sampling methods are insufficient for the flexible covalent ligands in the FAS system. Enhanced sampling methods that purely use the alchemical variable present in free energy calculations, \lambda , to enhance sampling include Hamiltonian replica exchange Jiang et al. 2018 and expanded ensemble Zhang et al. 2021 ; Åberg et al. 2004 . These methods to enhance sampling require that either the end states or the intermediate states have significantly different free energy barriers in configuration space. For the FAS examples proposed earlier, we are interested in the free energy of binding as a function of substrate chain length, which can vary by 2 carbon intervals from 4 to 16 carbons. Performing a single alchemical transformation from 4 to 16 carbons would not give information about the intermediate chain lengths, and performing iterative transformations would not result in sufficiently low FE barriers between alternate bound configurations at long chain lengths to converge sampling. Collective variable (CV) dependent methods, including metadynamics Min et al. 2007 ; Hsu et al. 2023 , umbrella sampling You et al. 2017 ; Torrie and Valleau 1977 , and weighted ensemble approaches Zwier et al. 2015 , represent another approach to enhanced sampling. However, they require significant prior knowledge about the conformational changes that enable transition between trapped metastable states. For the FAS examples, although we know that longer chains must be able to move between pockets, the number of degrees of freedom involved makes it challenging to identify relevant CVs that can describe such a transition involving all of the atoms. Intermediate motions along the proposed CV likely sample high-energy configurations, which decrease sampling efficiency. Even with an approximate path, these CV-dependent methods require prior knowledge of where any alternate pockets may be or significant trials to derive a more complete CV. We previously presented the enhanced sampling method, multiple topology replica exchange of expanded ensemble (MT-REXEE), which specifically targets enhancing sampling in flexible receptor-ligand binding systems Friedman et al. 2025 . In this work, we present improvements made to the MT-REXEE method through the introduction of a novel swapping algorithm, random range swapping, which increases the efficiency of the calculations and improves FE estimate convergence. This method works by enabling trajectory frames other than the last one in the iteration to be used