OpenMM 8: Molecular Dynamics Simulation with Machine Learning Potentials.

Machine learning plays an important and growing role in molecular simulation. The newest version of the OpenMM molecular dynamics toolkit introduces new features to support the use of machine learning potentials. Arbitrary PyTorch models can be added to a simulation and used to compute forces and energy. A higher-level interface allows users to easily model their molecules of interest with general purpose, pretrained potential functions. A collection of optimized CUDA kernels and custom PyTorch operations greatly improves the speed of simulations. We demonstrate these features in simulations of cyclin-dependent kinase 8 (CDK8) and the green fluorescent protein chromophore in water. Taken together, these features make it practical to use machine learning to improve the accuracy of simulations with only a modest increase in cost.

OpenMM 8: Molecular Dynamics Simulation with Machine Learning Potentials.

Machine learning plays an important and growing role in molecular simulation. The newest version of the OpenMM molecular dynamics toolkit introduces new features to support the use of machine learning potentials. Arbitrary PyTorch models can be added to a simulation and used to compute forces and energy. A higher-level interface allows users to easily model their molecules of interest with general purpose, pretrained potential functions. A collection of optimized CUDA kernels and custom PyTorch operations greatly improves the speed of simulations. We demonstrate these features on simulations of cyclin-dependent kinase 8 (CDK8) and the green fluorescent protein (GFP) chromophore in water. Taken together, these features make it practical to use machine learning to improve the accuracy of simulations at only a modest increase in cost.

Identifying and Overcoming the Sampling Challenges in Relative Binding Free Energy Calculations of a Model Protein:Protein Complex.

Relative alchemical binding free energy calculations are routinely used in drug discovery projects to optimize the affinity of small molecules for their drug targets. Alchemical methods can also be used to estimate the impact of amino acid mutations on protein:protein binding affinities, but these calculations can involve sampling challenges due to the complex networks of protein and water interactions frequently present in protein:protein interfaces. We investigate these challenges by extending a graphics processing unit (GPU)-accelerated open-source relative free energy calculation package (Perses) to predict the impact of amino acid mutations on protein:protein binding. Using the well-characterized model system barnase:barstar, we describe analyses for identifying and characterizing sampling problems in protein:protein relative free energy calculations. We find that mutations with sampling problems often involve charge-changes, and inadequate sampling can be attributed to slow degrees of freedom that are mutation-specific. We also explore the accuracy and efficiency of current state-of-the-art approaches─alchemical replica exchange and alchemical replica exchange with solute tempering─for overcoming relevant sampling problems. By employing sufficiently long simulations, we achieve accurate predictions (RMSE 1.61, 95% CI: [1.12, 2.11] kcal/mol), with 86% of estimates within 1 kcal/mol of the experimentally determined relative binding free energies and 100% of predictions correctly classifying the sign of the changes in binding free energies. Ultimately, we provide a model workflow for applying protein mutation free energy calculations to protein:protein complexes, and importantly, catalog the sampling challenges associated with these types of alchemical transformations. Our free open-source package (Perses) is based on OpenMM and is available at https://github.com/choderalab/perses.

Identifying and overcoming the sampling challenges in relative binding free energy calculations of a model protein:protein complex.

Relative alchemical binding free energy calculations are routinely used in drug discovery projects to optimize the affinity of small molecules for their drug targets. Alchemical methods can also be used to estimate the impact of amino acid mutations on protein:protein binding affinities, but these calculations can involve sampling challenges due to the complex networks of protein and water interactions frequently present in protein:protein interfaces. We investigate these challenges by extending a GPU-accelerated open-source relative free energy calculation package (Perses) to predict the impact of amino acid mutations on protein:protein binding. Using the well-characterized model system barnase:barstar, we describe analyses for identifying and characterizing sampling problems in protein:protein relative free energy calculations. We find that mutations with sampling problems often involve charge-changes, and inadequate sampling can be attributed to slow degrees of freedom that are mutation-specific. We also explore the accuracy and efficiency of current state-of-the-art approaches-alchemical replica exchange and alchemical replica exchange with solute tempering-for overcoming relevant sampling problems. By employing sufficiently long simulations, we achieve accurate predictions (RMSE 1.61, 95% CI: [1.12, 2.11] kcal/mol), with 86% of estimates within 1 kcal/mol of the experimentally-determined relative binding free energies and 100% of predictions correctly classifying the sign of the changes in binding free energies. Ultimately, we provide a model workflow for applying protein mutation free energy calculations to protein:protein complexes, and importantly, catalog the sampling challenges associated with these types of alchemical transformations. Our free open-source package (Perses) is based on OpenMM and available at https://github.com/choderalab/perses .

Long-term survival and second malignant tumor prediction in pediatric, adolescent, and young adult cancer survivors using Random Survival Forests: a SEER analysis.

Survival and second malignancy prediction models can aid clinical decision making. Most commonly, survival analysis studies are performed using traditional proportional hazards models, which require strong assumptions and can lead to biased estimates if violated. Therefore, this study aims to implement an alternative, machine learning (ML) model for survival analysis: Random Survival Forest (RSF). In this study, RSFs were built using the U.S. Surveillance Epidemiology and End Results to (1) predict 30-year survival in pediatric, adolescent, and young adult cancer survivors; and (2) predict risk and site of a second tumor within 30 years of the first tumor diagnosis in these age groups. The final RSF model for pediatric, adolescent, and young adult survival has an average Concordance index (C-index) of 92.9%, 94.2%, and 94.4% and average time-dependent area under the receiver operating characteristic curve (AUC) at 30-years since first diagnosis of 90.8%, 93.6%, 96.1% respectively. The final RSF model for pediatric, adolescent, and young adult second malignancy has an average C-index of 86.8%, 85.2%, and 88.6% and average time-dependent AUC at 30-years since first diagnosis of 76.5%, 88.1%, and 99.0% respectively. This study suggests the robustness and potential clinical value of ML models to alleviate physician burden by quickly identifying highest risk individuals.

End-to-end differentiable construction of molecular mechanics force fields.

Molecular mechanics (MM) potentials have long been a workhorse of computational chemistry. Leveraging accuracy and speed, these functional forms find use in a wide variety of applications in biomolecular modeling and drug discovery, from rapid virtual screening to detailed free energy calculations. Traditionally, MM potentials have relied on human-curated, inflexible, and poorly extensible discrete chemical perception rules (atom types) for applying parameters to small molecules or biopolymers, making it difficult to optimize both types and parameters to fit quantum chemical or physical property data. Here, we propose an alternative approach that uses graph neural networks to perceive chemical environments, producing continuous atom embeddings from which valence and nonbonded parameters can be predicted using invariance-preserving layers. Since all stages are built from smooth neural functions, the entire process-spanning chemical perception to parameter assignment-is modular and end-to-end differentiable with respect to model parameters, allowing new force fields to be easily constructed, extended, and applied to arbitrary molecules. We show that this approach is not only sufficiently expressive to reproduce legacy atom types, but that it can learn to accurately reproduce and extend existing molecular mechanics force fields. Trained with arbitrary loss functions, it can construct entirely new force fields self-consistently applicable to both biopolymers and small molecules directly from quantum chemical calculations, with superior fidelity than traditional atom or parameter typing schemes. When adapted to simultaneously fit partial charge models, espaloma delivers high-quality partial atomic charges orders of magnitude faster than current best-practices with low inaccuracy. When trained on the same quantum chemical small molecule dataset used to parameterize the Open Force Field ("Parsley") openff-1.2.0 small molecule force field augmented with a peptide dataset, the resulting espaloma model shows superior accuracy vis-á-vis experiments in computing relative alchemical free energy calculations for a popular benchmark. This approach is implemented in the free and open source package espaloma, available at https://github.com/choderalab/espaloma.

SARS-CoV-2 RBD antibodies that maximize breadth and resistance to escape.

An ideal therapeutic anti-SARS-CoV-2 antibody would resist viral escape1-3, have activity against diverse sarbecoviruses4-7, and be highly protective through viral neutralization8-11 and effector functions12,13. Understanding how these properties relate to each other and vary across epitopes would aid the development of therapeutic antibodies and guide vaccine design. Here we comprehensively characterize escape, breadth and potency across a panel of SARS-CoV-2 antibodies targeting the receptor-binding domain (RBD). Despite a trade-off between in vitro neutralization potency and breadth of sarbecovirus binding, we identify neutralizing antibodies with exceptional sarbecovirus breadth and a corresponding resistance to SARS-CoV-2 escape. One of these antibodies, S2H97, binds with high affinity across all sarbecovirus clades to a cryptic epitope and prophylactically protects hamsters from viral challenge. Antibodies that target the angiotensin-converting enzyme 2 (ACE2) receptor-binding motif (RBM) typically have poor breadth and are readily escaped by mutations despite high neutralization potency. Nevertheless, we also characterize a potent RBM antibody (S2E128) with breadth across sarbecoviruses related to SARS-CoV-2 and a high barrier to viral escape. These data highlight principles underlying variation in escape, breadth and potency among antibodies that target the RBD, and identify epitopes and features to prioritize for therapeutic development against the current and potential future pandemics.

Antibodies to the SARS-CoV-2 receptor-binding domain that maximize breadth and resistance to viral escape.

An ideal anti-SARS-CoV-2 antibody would resist viral escape 1-3 , have activity against diverse SARS-related coronaviruses 4-7 , and be highly protective through viral neutralization 8-11 and effector functions 12,13 . Understanding how these properties relate to each other and vary across epitopes would aid development of antibody therapeutics and guide vaccine design. Here, we comprehensively characterize escape, breadth, and potency across a panel of SARS-CoV-2 antibodies targeting the receptor-binding domain (RBD), including S309 4 , the parental antibody of the late-stage clinical antibody VIR-7831. We observe a tradeoff between SARS-CoV-2 in vitro neutralization potency and breadth of binding across SARS-related coronaviruses. Nevertheless, we identify several neutralizing antibodies with exceptional breadth and resistance to escape, including a new antibody (S2H97) that binds with high affinity to all SARS-related coronavirus clades via a unique RBD epitope centered on residue E516. S2H97 and other escape-resistant antibodies have high binding affinity and target functionally constrained RBD residues. We find that antibodies targeting the ACE2 receptor binding motif (RBM) typically have poor breadth and are readily escaped by mutations despite high neutralization potency, but we identify one potent RBM antibody (S2E12) with breadth across sarbecoviruses closely related to SARS-CoV-2 and with a high barrier to viral escape. These data highlight functional diversity among antibodies targeting the RBD and identify epitopes and features to prioritize for antibody and vaccine development against the current and potential future pandemics.

Circulating SARS-CoV-2 spike N439K variants maintain fitness while evading antibody-mediated immunity.

SARS-CoV-2 can mutate and evade immunity, with consequences for efficacy of emerging vaccines and antibody therapeutics. Here, we demonstrate that the immunodominant SARS-CoV-2 spike (S) receptor binding motif (RBM) is a highly variable region of S and provide epidemiological, clinical, and molecular characterization of a prevalent, sentinel RBM mutation, N439K. We demonstrate N439K S protein has enhanced binding affinity to the hACE2 receptor, and N439K viruses have similar in vitro replication fitness and cause infections with similar clinical outcomes as compared to wild type. We show the N439K mutation confers resistance against several neutralizing monoclonal antibodies, including one authorized for emergency use by the US Food and Drug Administration (FDA), and reduces the activity of some polyclonal sera from persons recovered from infection. Immune evasion mutations that maintain virulence and fitness such as N439K can emerge within SARS-CoV-2 S, highlighting the need for ongoing molecular surveillance to guide development and usage of vaccines and therapeutics.

Heparan Sulfate Proteoglycan Sulfation Regulates Uterine Differentiation and Signaling During Embryo Implantation.

To prepare for embryo implantation, the uterus must undergo a series of reciprocal interactions between the uterine epithelium and the underlying stroma, which are orchestrated by ovarian hormones. During this process, multiple signaling pathways are activated to direct cell proliferation and differentiation, which render the uterus receptive to the implanting blastocysts. One important modulator of these signaling pathways is the cell surface and extracellular matrix macromolecules, heparan sulfate proteoglycans (HSPGs). HSPGs play crucial roles in signal transduction by regulating morphogen transport and ligand binding. In this study, we examine the role of HSPG sulfation in regulating uterine receptivity by conditionally deleting the N-deacetylase/N-sulfotransferase (NDST) 1 gene (Ndst1) in the mouse uterus using the Pgr-Cre driver, on an Ndst2- and Ndst3-null genetic background. Although development of the female reproductive tract and subsequent ovarian function appear normal in Ndst triple-knockout females, they are infertile due to implantation defects. Embryo attachment appears to occur but the uterine epithelium at the site of implantation persists rather than disintegrates in the mutant. Uterine epithelial cells continued to proliferate past day 4 of pregnancy, accompanied by elevated Fgf2 and Fgf9 expression, whereas uterine stroma failed to undergo decidualization, as evidenced by lack of Bmp2 induction. Despite normal Indian hedgehog expression, transcripts of Ptch1 and Gli1, both components as well as targets of the hedgehog (Hh) pathway, were detected only in the subepithelial stroma, indicating altered Hh signaling in the mutant uterus. Taken together, these data implicate an essential role for HSPGs in modulating signal transduction during mouse implantation.

Proteome-Scale Human Interactomics.

Cellular functions are mediated by complex interactome networks of physical, biochemical, and functional interactions between DNA sequences, RNA molecules, proteins, lipids, and small metabolites. A thorough understanding of cellular organization requires accurate and relatively complete models of interactome networks at proteome scale. The recent publication of four human protein-protein interaction (PPI) maps represents a technological breakthrough and an unprecedented resource for the scientific community, heralding a new era of proteome-scale human interactomics. Our knowledge gained from these and complementary studies provides fresh insights into the opportunities and challenges when analyzing systematically generated interactome data, defines a clear roadmap towards the generation of a first reference interactome, and reveals new perspectives on the organization of cellular life.

Homeodomain Transcription Factor Msx-2 Regulates Uterine Progenitor Cell Response to Diethylstilbestrol.

The fate of mouse uterine epithelial progenitor cells is determined between postnatal days 5 to 7. Around this critical time window, exposure to an endocrine disruptor, diethylstilbestrol (DES), can profoundly alter uterine cytodifferentiation. We have shown previously that a homeo domain transcription factor MSX-2 plays an important role in DES-responsiveness in the female reproductive tract (FRT). Mutant FRTs exhibited a much more severe phenotype when treated with DES, accompanied by gene expression changes that are dependent on Msx2. To better understand the role that MSX-2 plays in uterine response to DES, we performed global gene expression profiling experiment in mice lacking Msx2 By comparing this result to our previously published microarray data performed on wild-type mice, we extracted common and differentially regulated genes in the two genotypes. In so doing, we identified potential downstream targets of MSX-2, as well as genes whose regulation by DES is modulated through MSX-2. Discovery of these genes will lead to a better understanding of how DES, and possibly other endocrine disruptors, affects reproductive organ development.