Biology subjects

Bronstein, M.

Publications and source records attributed to Bronstein, M..

11 recordsLinked to original sources

bioRxiv2026

A Unified 3D Generative Model for Synthesizable Structure-Based Drug Design

Traditional screening-based drug discovery is inherently limited by the astronomical scale of the chemical space. Generative modelling offers a compelling alternative to the classical search paradigm and enables rational, bottom-up design of novel and target-specific small molecules. However, its impact has been hampered by challenges in synthetic accessibility of the designed compounds and lack of large-scale experimental validation. Here, we introduce LDDM (Large Drug Discovery Model), a generative framework that supports a range of drug discovery tasks, including constrained and unconstrained docking, fragment linking and growing, and de novo design. We further introduce a programmable design algorithm that enables accurate design of synthetically accessible compounds satisfying various fine-grained objectives. We experimentally validated the designed or optimised ligands for five therapeutically relevant protein targets. In all cases, LDDM achieved high success rates, allowing us to identify molecules with confirmed binding affinity while synthesizing only a small number of generated compounds. The best designs were structurally characterised through NMR spectroscopy and X-ray crystallography, demonstrating high prediction accuracy. Overall, LDDM provides a scalable and flexible platform for the rapid and tailored design of small molecules and non-natural peptides for therapeutic applications.

bioinformatics↗

bioRxiv2025

FragmentScope - exploring the fragment space with learned surface representations

Exploring fragment chemical space for ligand design remains a major challenge in early stage drug discovery. This task is particularly challenging due to the small size, low specificity, and weak binding affinities of low molecular weight (MW) fragments. We present FragmentScope, a computational pipeline that uses learned protein surface fingerprints to guide fragment placement and small molecule generation. By using a contrastive learning model trained on protein-ligand interactions, we built a database of surface-fragment pairs, which enables fast and accurate fragment placement in the target protein pocket. We demonstrate its effectiveness on benchmark datasets, achieving robust placement accuracy. FragmentScope also enables the design of small molecules based on the predicted fragments ensuring synthetic accessibility. We experimentally validated Fragmentscope across 5 different targets with binding assays and structural characterization. Our approach shows high success rates in fragment discovery and yielded promising leads for designed ligands. FragmentScope offers a scalable, structure-guided approach for narrowing chemical space and identifying prospective scaffolds, accelerating the early stages of small molecule design.

bioinformatics↗

bioRxiv2024

PINDER: The protein interaction dataset and evaluation resource

Protein-protein interactions (PPIs) are fundamental to understanding biological processes and play a key role in therapeutic advancements. As deep-learning docking methods for PPIs gain traction, benchmarking protocols and datasets tailored for effective training and evaluation of their generalization capabilities and performance across real-world scenarios become imperative. Aiming to overcome limitations of existing approaches, we introduce PINDER, a comprehensive annotated dataset that uses structural clustering to derive non-redundant interface-based data splits and includes holo (bound), apo (unbound), and computationally predicted structures. PINDER consists of 2,319,564 dimeric PPI systems (and up to 25 million augmented PPIs) and 1,955 high-quality test PPIs with interface data leakage removed. Additionally, PINDER provides a test subset with 180 dimers for comparison to AlphaFold-Multimer without any interface leakage with respect to its training set. Unsurprisingly, the PINDER benchmark reveals that the performance of existing docking models is highly overestimated when evaluated on leaky test sets. Most importantly, by retraining DiffDock-PP on PINDER interface-clustered splits, we show that interface cluster-based sampling of the training split, along with the diverse and less leaky validation split, leads to strong generalization improvements.

bioinformatics↗

bioRxiv2024

PLINDER: The protein-ligand interactions dataset and evaluation resource

Protein-ligand interactions (PLI) are foundational to small molecule drug design. With computational methods striving towards experimental accuracy, there is a critical demand for a well-curated and diverse PLI dataset. Existing datasets are often limited in size and diversity, and commonly used evaluation sets suffer from training information leakage, hindering the realistic assessment of method generalization capabilities. To address these shortcomings, we present PLIN-DER, the largest and most annotated dataset to date, comprising 449,383 PLI systems, each with over 500 annotations, similarity metrics at protein, pocket, interaction and ligand levels, and paired unbound (apo) and predicted structures. We propose an approach to generate training and evaluation splits that minimizes task-specific leakage and maximizes test set quality, and compare the resulting performance of DiffDock when retrained with different kinds of splits.

biochemistry↗

bioRxiv2024

Targeting protein-ligand neosurfaces using a generalizable deep learning approach

Molecular recognition events between proteins drive biological processes in living systems. However, higher levels of mechanistic regulation have emerged, where protein-protein interactions are conditioned to small molecules. Here, we present a computational strategy for the design of proteins that target neosurfaces, i.e. surfaces arising from protein-ligand complexes. To do so, we leveraged a deep learning approach based on learned molecular surface representations and experimentally validated binders against three drug-bound protein complexes. Remarkably, surface fingerprints trained only on proteins can be applied to neosurfaces emerging from small molecules, serving as a powerful demonstration of generalizability that is uncommon in deep learning approaches. The designed chemically-induced protein interactions hold the potential to expand the sensing repertoire and the assembly of new synthetic pathways in engineered cells.

biochemistry↗

bioRxiv2024

Combinatorial prediction of therapeutic perturbations using causally-inspired neural networks

Phenotype-driven approaches identify disease-counteracting compounds by analyzing the phenotypic signatures that distinguish diseased from healthy states. Here, we introduce PDGrapher, a causally inspired graph neural network (GNN) model that predicts combinatorial perturbagens (sets of therapeutic targets) capable of reversing disease phenotypes. Unlike methods that learn how perturbations alter phenotypes, PDGrapher solves the inverse problem of directly predicting the perturbagens needed to achieve a desired response by embedding disease cell states into networks, learning a latent representation of these states, and identifying optimal combinatorial perturbations. In experiments in nine cell lines with chemical perturbations, PDGrapher identifies effective perturbagens in more testing samples than competing methods. It also demonstrates competitive performance on ten genetic perturbation datasets. An advantage of PDGrapher is its direct prediction, in contrast to the indirect and computationally intensive approach traditionally used in phenotype-driven models. This approach accelerates training by up to 25 times compared to existing methods, providing a fast approach for identifying therapeutic perturbations and advancing phenotype-driven drug discovery.

bioinformatics↗

bioRxiv2023

Exploring "dark matter" protein folds using deep learning

De novo protein design aims to explore uncharted sequence-and structure areas to generate novel proteins that have not been sampled by evolution. One of the main challenges in de novo design involves crafting "designable" structural templates that can guide the sequence search towards adopting the target structures. Here, we present an approach to learn patterns of protein structure based on a convolutional variational autoencoder, dubbed Genesis. We coupled Genesis with trRosetta to design sequences for a set of protein folds and found that Genesis is capable of reconstructing native-like distance-and angle distributions for five native folds and three novel, so-called "dark-matter" folds as a demonstration of generalizability. We used a high-throughput assay to characterize protease resistance of the designs, obtaining encouraging success rates for folded proteins and further biochemically characterized folded designs. The Genesis framework enables the exploration of the protein sequence and fold space within minutes and is not bound to specific protein topologies. Our approach addresses the backbone designability problem, showing that structural patterns in proteins can be efficiently learned by small neural networks and could ultimately contribute to the de novo design of proteins with new functions.

bioinformatics↗

bioRxiv2022

De novo design of site-specific protein interactions with learned surface fingerprints

Physical interactions between proteins are essential for most biological processes governing life. However, the molecular determinants of such interactions have been challenging to understand, even as genomic, proteomic, and structural data grows. This knowledge gap has been a major obstacle for the comprehensive understanding of cellular protein-protein interaction (PPI) networks and for the de novo design of protein binders that are crucial for synthetic biology and translational applications. We exploit a geometric deep learning framework operating on protein surfaces that generates fingerprints to describe geometric and chemical features critical to drive PPIs. We hypothesized these fingerprints capture the key aspects of molecular recognition that represent a new paradigm in the computational design of novel protein interactions. As a proof-of-principle, we computationally designed several de novo protein binders to engage four protein targets: SARS-CoV-2 spike, PD-1, PD-L1, and CTLA-4. Several designs were experimentally optimized while others were purely generated in silico, reaching nanomolar affinity with structural and mutational characterization showing highly accurate predictions. Overall, our surface-centric approach captures the physical and chemical determinants of molecular recognition, enabling a novel approach for the de novo design of protein interactions and, more broadly, of artificial proteins with function.

bioengineering↗

bioRxiv2022

Decoding Surface Fingerprints for Protein-Ligand Interactions

AO_SCPLOWBSTRACTC_SCPLOWSmall molecules have been the preferred modality for drug development and therapeutic interventions. This molecular format presents a number of advantages, e.g. long half-lives and cell permeability, making it possible to access a wide range of therapeutic targets. However, finding small molecules that engage "hard-to-drug" protein targets specifically and potently remains an arduous process, requiring experimental screening of extensive compound libraries to identify candidate leads. The search continues with further optimization of compound leads to meet the required potency and toxicity thresholds for clinical applications. Here, we propose a new computational workflow for high-throughput fragment-based screening and binding affinity prediction where we leverage the available protein-ligand complex structures using a state-of-the-art protein surface embedding framework (dMaSIF). We developed a tool capable of finding suitable ligands and fragments for a given protein pocket solely based on protein surface descriptors, that capture chemical and geometric features of the target pocket. The identified fragments can be further combined into novel ligands. Using the structural data, our ligand discovery pipeline learns the signatures of interactions between surface patches and small pharmacophores. On a query target pocket, the algorithm matches known target pockets and returns either potential ligands or identifies multiple ligand fragments in the binding site. Our binding affinity predictor is capable of predicting the affinity of a given protein-ligand pair, requiring only limited information about the ligand pose. This enables screening without the costly step of first docking candidate molecules. Our framework will facilitate the design of ligands based on the targets surface information. It may significantly reduce the experimental screening load and ultimately reveal novel chemical compounds for targeting challenging proteins.

bioinformatics↗

bioRxiv2021

Human nasal and lung tissues infected ex vivo with SARS-CoV-2 provide insights into differential tissue-specific and virus-specific innate immune responses in the upper and lower respiratory tract

The nasal-mucosa constitutes the primary entry site for respiratory viruses including SARS-CoV-2. While the imbalanced innate immune response of end-stage COVID-19 has been extensively studied, the earliest stages of SARS-CoV-2 infection at the mucosal entry site have remained unexplored. Here we employed SARS-CoV-2 and influenza virus infection in native multi-cell-type human nasal turbinate and lung tissues ex vivo, coupled with genome-wide transcriptional analysis, to investigate viral susceptibility and early patterns of local-mucosal innate immune response in the authentic milieu of the human respiratory tract. SARS-CoV-2 productively infected the nasal turbinate tissues, predominantly targeting respiratory epithelial cells, with rapid increase in tissue-associated viral sub-genomic mRNA, and secretion of infectious viral progeny. Importantly, SARS-CoV-2 infection triggered robust antiviral and inflammatory innate immune responses in the nasal mucosa. The upregulation of interferon stimulated genes, cytokines and chemokines, related to interferon signaling and immune-cell activation pathways, was broader than that triggered by influenza virus infection. Conversely, lung tissues exhibited a restricted innate immune response to SARS-CoV-2, with a conspicuous lack of type I and III interferon upregulation, contrasting with their vigorous innate immune response to influenza virus. Our findings reveal differential tissue-specific innate immune responses in the upper and lower respiratory tract, that are distinct to SARS-CoV-2. The studies shed light on the role of the nasal-mucosa in active viral transmission and immune defense, implying a window of opportunity for early interventions, whereas the restricted innate immune response in early-SARS-CoV-2-infected lung tissues could underlie the unique uncontrolled late-phase lung damage of advanced COVID-19. IMPORTANCEIn order to reduce the late-phase morbidity and mortality of COVID-19, there is a need to better understand and target the earliest stages of SARS-CoV-2 infection in the human respiratory tract. Here we have studied the initial steps of SARS-CoV-2 infection and the consequent innate immune responses within the natural multicellular complexity of human nasal-mucosal and lung tissues. Comparing the global innate response patterns of nasal and lung tissues, infected in parallel with SARS-CoV-2 and influenza virus, we have revealed distinct virus-host interactions in the upper and lower respiratory tract, which could determine the outcome and unique pathogenesis of SARS-CoV-2 infection. Studies in the nasal-mucosal infection model can be employed to assess the impact of viral evolutionary changes, and evaluate new therapeutic and preventive measures against SARS-CoV-2 and other human respiratory pathogens.

microbiology↗

bioRxiv2020

Fast end-to-end learning on protein surfaces

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.

bioinformatics↗