Generative models see increasing use in computer-aided drug design. However, while performing well at capturing distributions of molecular motifs, they often produce synthetically inaccessible molecules. To address this, we introduce SynFlowNet, a GFlowNet model whose action space uses chemical reactions and buyable reactants to sequentially build new molecules. By incorporating forward synthesis as an explicit constraint of the generative mechanism, we aim at bridging the gap between in silico molecular generation and real world synthesis capabilities. We evaluate our approach using synthetic accessibility scores and an independent retrosynthesis tool to assess the synthesizability of our compounds, and motivate the choice of GFlowNets through considerable improvement in sample diversity compared to baselines. Additionally, we identify challenges with reaction encodings that can complicate traversal of the MDP in the backward direction. To address this, we introduce various strategies for learning the GFlowNet backward policy and thus demonstrate how additional constraints can be integrated into the GFlowNet MDP framework. This approach enables our model to successfully identify synthesis pathways for previously unseen molecules.

Machine learning for protein modeling faces significant challenges due to proteins' inherently dynamic nature, yet most graph-based machine learning methods rely solely on static structural information. Recently, the growing availability of molecular dynamics trajectories provides new opportunities for understanding the dynamic behavior of proteins; however, computational methods for utilizing this dynamic information remain limited. We propose a novel graph representation that integrates both static structural information and dynamic correlations from molecular dynamics trajectories, enabling more comprehensive modeling of proteins. By applying relational graph neural networks (RGNNs) to process this heterogeneous representation, we demonstrate significant improvements over structure-based approaches across three distinct tasks: atomic adaptability prediction, binding site detection, and binding affinity prediction. Our results validate that combining static and dynamic information provides complementary signals for understanding protein-ligand interactions, offering new possibilities for drug design and structural biology applications.

Retrosynthesis planning is a fundamental challenge in chemistry which aims at designing multi-step reaction pathways from commercially available starting materials to a target molecule. Each step in multi-step retrosynthesis planning requires accurate prediction of possible precursor molecules given the target molecule and confidence estimates to guide heuristic search algorithms. We model single-step retrosynthesis as a distribution learning problem in a discrete state space. First, we introduce the Markov Bridge Model, a generative framework aimed to approximate the dependency between two intractable discrete distributions accessible via a finite sample of coupled data points. Our framework is based on the concept of a Markov bridge, a Markov process pinned at its endpoints. Unlike diffusion-based methods, our Markov Bridge Model does not need a tractable noise distribution as a sampling proxy and directly operates on the input product molecules as samples from the intractable prior distribution. We then address the retrosynthesis planning problem with our novel framework and introduce RetroBridge, a template-free retrosynthesis modeling approach that achieves state-of-the-art results on standard evaluation benchmarks.

Proteins are typically represented by discrete atomic coordinates providing an accessible framework to describe different conformations. However, in some fields proteins are more accurately represented as near-continuous surfaces, as these are imprinted with geometric (shape) and chemical (electrostatics) features of the underlying protein structure. Protein surfaces are dependent on their chemical composition and, ultimately determine protein function, acting as the interface that engages in interactions with other molecules. In the past, such representations were utilized to compare protein structures on global and local scales and have shed light on functional properties of proteins. Here we describe RosettaSurf, a surface-centric computational design protocol, that focuses on the molecular surface shape and electrostatic properties as means for protein engineering, offering a unique approach for the design of proteins and their functions. The RosettaSurf protocol combines the explicit optimization of molecular surface features with a global scoring function during the sequence design process, diverging from the typical design approaches that rely solely on an energy scoring function. With this computational approach, we attempt to address a fundamental problem in protein design related to the design of functional sites in proteins, even when structurally similar templates are absent in the characterized structural repertoire. Surface-centric design exploits the premise that molecular surfaces are, to a certain extent, independent of the underlying sequence and backbone configuration, meaning that different sequences in different proteins may present similar surfaces. We benchmarked RosettaSurf on various sequence recovery datasets and showcased its design capabilities by generating epitope mimics that were biochemically validated. Overall, our results indicate that the explicit optimization of surface features may lead to new routes for the design of functional proteins.

Proteins’ biological functions are defined by the geometric and chemical structure of their 3D molecular surfaces. Recent works have shown that geometric deep learning can be used on mesh-based representations of proteins to identify potential functional sites, such as binding targets for potential drugs. Unfortunately though, the use of meshes as the underlying representation for protein structure has multiple drawbacks including the need to pre-compute the input features and mesh connectivities. This becomes a bottleneck for many important tasks in protein science.In this paper, we present a new framework for deep learning on protein structures that addresses these limitations. Among the key advantages of our method are the computation and sampling of the molecular surface on-the-fly from the underlying atomic point cloud and a novel efficient geometric convolutional layer. As a result, we are able to process large collections of proteins in an end-to-end fashion, taking as the sole input the raw 3D coordinates and chemical types of their atoms, eliminating the need for any hand-crafted pre-computed features.To showcase the performance of our approach, we test it on two tasks in the field of protein structural bioinformatics: the identification of interaction sites and the prediction of protein-protein interactions. On both tasks, we achieve state-of-the-art performance with much faster run times and fewer parameters than previous models. These results will considerably ease the deployment of deep learning methods in protein science and open the door for end-to-end differentiable approaches in protein modeling tasks such as function prediction and design.

BACKGROUND: Large-scale datasets of protein structures and sequences are becoming ubiquitous in many domains of biological research. Experimental approaches and computational modelling methods are generating biological data at an unprecedented rate. The detailed analysis of structure-sequence relationships is critical to unveil governing principles of protein folding, stability and function. Computational protein design (CPD) has emerged as an important structure-based approach to engineer proteins for novel functions. Generally, CPD workflows rely on the generation of large numbers of structural models to search for the optimal structure-sequence configurations. As such, an important step of the CPD process is the selection of a small subset of sequences to be experimentally characterized. Given the limitations of current CPD scoring functions, multi-step design protocols and elaborated analysis of the decoy populations have become essential for the selection of sequences for experimental characterization and the success of CPD strategies. RESULTS: Here, we present the rstoolbox, a Python library for the analysis of large-scale structural data tailored for CPD applications. rstoolbox is oriented towards both CPD software users and developers, being easily integrated in analysis workflows. For users, it offers the ability to profile and select decoy sets, which may guide multi-step design protocols or for follow-up experimental characterization. rstoolbox provides intuitive solutions for the visualization of large sequence/structure datasets (e.g. logo plots and heatmaps) and facilitates the analysis of experimental data obtained through traditional biochemical techniques (e.g. circular dichroism and surface plasmon resonance) and high-throughput sequencing. For CPD software developers, it provides a framework to easily benchmark and compare different CPD approaches. Here, we showcase the rstoolbox in both types of applications. CONCLUSIONS: rstoolbox is a library for the evaluation of protein structures datasets tailored for CPD data. It provides interactive access through seamless integration with IPython, while still being suitable for high-performance computing. In addition to its functionalities for data analysis and graphical representation, the inclusion of rstoolbox in protein design pipelines will allow to easily standardize the selection of design candidates, as well as, to improve the overall reproducibility and robustness of CPD selection processes.

The robust computational design of functional proteins has the potential to deeply impact translational research and broaden our understanding of the determinants of protein function and stability. The low success rates of computational design protocols and the extensive in vitro optimization often required, highlight the challenge of designing proteins that perform essential biochemical functions, such as binding or catalysis. One of the most simplistic approaches for the design of function is to adopt functional motifs in naturally occurring proteins and transplant them to computationally designed proteins. The structural complexity of the functional motif largely determines how readily one can find host protein structures that are "designable", meaning that are likely to present the functional motif in the desired conformation. One promising route to enhance the "designability" of protein structures is to allow backbone flexibility. Here, we present a computational approach that couples conformational folding with sequence design to embed functional motifs into heterologous proteins-Rosetta Functional Folding and Design (FunFolDes). We performed extensive computational benchmarks, where we observed that the enforcement of functional requirements resulted in designs distant from the global energetic minimum of the protein. An observation consistent with several experimental studies that have revealed function-stability tradeoffs. To test the design capabilities of FunFolDes we transplanted two viral epitopes into distant structural templates including one de novo "functionless" fold, which represent two typical challenges where the designability problem arises. The designed proteins were experimentally characterized showing high binding affinities to monoclonal antibodies, making them valuable candidates for vaccine design endeavors. Overall, we present an accessible strategy to repurpose old protein folds for new functions. This may lead to important improvements on the computational design of proteins, with structurally complex functional sites, that can perform elaborate biochemical functions related to binding and catalysis.