Joe G. Greener

Force fields for molecular dynamics are usually developed manually, limiting their transferability and making systematic exploration of functional forms challenging.

Structural bioinformatics aims to answer biological questions by considering biomolecular structures at scale. We now have accurate predictions and need to ask what we can learn from them.

The next generation of force fields for molecular dynamics will be developed using a wealth of data. Training systematically with experimental data remains a challenge, however, especially for machine-learning potentials. Differentiable molecular simulation calculates gradients of observables with respect to parameters through molecular dynamics trajectories. Here, we improve this approach by explicitly calculating gradients using a reverse-time simulation with effectively constant memory cost and a computation count similar to the forward simulation. The method is applied to learn all-atom water and gas diffusion models with different functional forms and to train a machine-learning potential for diamond from scratch. Comparison to ensemble reweighting indicates that reversible simulation can provide more accurate gradients and train to match time-dependent observables.

A force field as accurate as quantum mechanics (QMs) and as fast as molecular mechanics (MMs), with which one can simulate a biomolecular system efficiently enough and meaningfully enough to get quantitative insights, is among the most ardent dreams of biophysicists—a dream, nevertheless, not to be fulfilled any time soon. Machine learning force fields (MLFFs) represent a meaningful endeavor in this direction, where differentiable neural functions are parametrized to fit ab initio energies and forces through automatic differentiation. We argue that, as of now, the utility of the MLFF models is no longer bottlenecked by accuracy but primarily by their speed, as well as stability and generalizability—many recent variants, on limited chemical spaces, have long surpassed the chemical accuracy of 1 kcal/mol—the empirical threshold beyond which realistic chemical predictions are possible—though still magnitudes slower than MM. Hoping to kindle exploration and design of faster, albeit perhaps slightly less accurate MLFFs, in this review, we focus our attention on the technical design space (the speed-accuracy trade-off) between MM and ML force fields. After a brief review of the building blocks (from a machine learning-centric point of view) of force fields of either kind, we discuss the desired properties and challenges now faced by the force field development community, survey the efforts to make MM force fields more accurate and ML force fields faster, and envision what the next generation of MLFF might look like.

: Comparing and searching protein structures independent of primary sequence has proved useful for remote homology detection, function annotation, and protein classification. Fast and accurate methods to search with structures will be essential to make use of the vast databases that have recently become available, in the same way that fast protein sequence searching underpins much of bioinformatics. We train a simple graph neural network using supervised contrastive learning to learn a low-dimensional embedding of protein domains. Availability and implementation: The method, called Progres, is available as software at https://github.com/greener-group/progres and as a web server at https://progres.mrc-lmb.cam.ac.uk. It has accuracy comparable to the best current methods and can search the AlphaFold database TED domains in a 10th of a second per query on CPU.

During mitosis, unattached kinetochores trigger the spindle assembly checkpoint by promoting assembly of the mitotic checkpoint complex, a heterotetramer comprising Mad2, Cdc20, BubR1 and Bub3. Critical to this process is the kinetochore-mediated catalysis of an intrinsically slow conformational conversion of Mad2 from an open (O-Mad2) inactive state to a closed (C-Mad2) active state bound to Cdc20. These Mad2 conformational changes involve substantial remodelling of the N-terminal β1 strand and C-terminal β7/β8 hairpin. In vitro , the Mad2- interaction motif (MIM) of Cdc20 (Cdc20 ^MIM ) triggers rapid conversion of O- to C-Mad2, effectively removing the kinetic barrier for MCC assembly. How Cdc20 ^MIM directly induces Mad2 conversion remains unclear. In this study we demonstrate that the Cdc20 ^MIM -binding site is inaccessible in O-Mad2. Time-resolved NMR and molecular dynamics simulations show how Mad2 conversion involves sequential conformational changes of flexible structural elements in O-Mad2, orchestrated by Cdc20 ^MIM . Conversion is initiated by the β7/β8 hairpin of O-Mad2 transiently unfolding to expose a nascent Cdc20 ^MIM -binding site. Engagement of Cdc20 ^MIM to this site promotes release of the β1 strand. We propose that initial conformational changes of the β7/β8 hairpin allows binding of Cdc20 ^MIM to a transient intermediate state of Mad2, thereby lowering the kinetic barrier to Mad2 conversion.

The a99SB- disp force field and GBNeck2 implicit solvent model are improved to better describe disordered proteins. 5 ns differentiable molecular simulations are used to jointly optimise 108 parameters to better match explicit solvent trajectories.

Implicit solvent force fields are computationally efficient but can be unsuitable for running molecular dynamics on disordered proteins. Here I improve the a99SB- disp force field and the GBNeck2 implicit solvent model to better describe disordered proteins. Differentiable molecular simulations with 5 ns trajectories are used to jointly optimise 108 parameters to better match explicit solvent trajectories. Simulations with the improved force field better reproduce the radius of gyration and secondary structure content seen in experiments, whilst showing slightly degraded performance on folded proteins and protein complexes. The force field, called GB99dms, reproduces the results of a small molecule binding study and improves agreement to experiment for the aggregation of amyloid peptides. GB99dms, which can be used in OpenMM, is available at https://github.com/greener-group/GB99dms . This work is the first to show that gradients can be obtained directly from nanosecond-length differentiable simulations of biomolecules and highlights the effectiveness of this approach to training whole force fields to match desired properties.

This chapter documents and theorizes the connection between contemporary dynamics of neo-imperialism and social orders defined by crime, repression, and violence. In opposition to Southern criminology framings, we argue for a materialist understanding of neo-imperialism, placing processes of wealth transference and extraction from Global South to North as central to the global system of inequality. This set of economic relations of exploitation sits within complex political relations that regulate and legitimize business practices. Maintaining unequal wealth accumulation from the South in the face of crisis, critique, and resistance entails the constant deployment of strategies of consent and repression. After this, we develop two discussions that theoretically ground the intimate connection between neo-colonialism and crime, violence, and state repression in the Global South. Firstly, we argue that corporate accumulation strategies based on the redistribution of wealth from South to North frequently depend on the institutionalization of de-regulated corporate permission which is associated with a range of harms. Secondly, we argue that through trade dynamics and economic policy, often enforced by multilateral institutions, poverty in the South is a structural outcome leading to constant problems of social disorder as informal economies and reactionary state violence take hold in the face of insecurity.