Biology subjects

Crivelli-Decker, J.

Publications and source records attributed to Crivelli-Decker, J..

4 recordsLinked to original sources

bioRxiv2026

On improving experimental binding affinity predictions with synthetic data

The success of deep learning binding affinity prediction models depends critically on expanding experimental data with reliable synthetic data. We extend the Structurally Augmented IC50 Repository (SAIR) with {approx}80K absolute free energy perturbation (AFEP) calculations and present two distinct data splits, SAIR-FEP and SAIR-OOD (out-of-distribution), to simulate realistic drug discovery scenarios. We compare sequence-based proteochemometric (PCM) models and state-of-the-art, structure-based deep learning models and demonstrate that PCM models can be enhanced by physics-based descriptors. While structure-based deep learning methods capture finer geometric detail, their performance is highly sensitive to the input structure. By filtering for high-confidence, co-folded complexes, we show that the performance improves predictably, whereas training on all complexes blindly does not yield performance gains. Finally, using the SAIR-OOD split, we demonstrate that simultaneous training on synthetic and experimental data improves performance on publicly available, experimental benchmarks. These results provide a clear strategy for using synthetic data to advance experimental binding affinity predictions.

molecular biology↗

bioRxiv2025

SAIR: Enabling deep learning for protein-ligand lnteractions with a synthetic structural dataset

AO_SCPLOWBSTRACTC_SCPLOWAccurate prediction of protein-ligand binding affinities remains a cornerstone problem in drug discovery. While binding affinity is inherently dictated by the 3D structure and dynamics of protein-ligand complexes, current deep learning approaches are limited by the lack of high-quality experimental structures with annotated binding affinities. To address this limitation, we introduce the Struc-turally Augmented IC50 Repository (SAIR), the largest publicly available dataset of protein-ligand 3D structures with associated activity data. The dataset com-prises 5, 244, 285 structures across 1, 048, 857 unique protein-ligand systems, cu-rated from the ChEMBL and BindingDB databases, which were then computa-tionally folded using the Boltz-1x model. We provide a comprehensive charac-terization of the dataset, including distributional statistics of proteins and ligands, and evaluate the structural fidelity of the folded complexes using PoseBusters. Our analysis reveals that approximately 3% of structures exhibit physical anoma-lies, predominantly related to internal energy violations. As an initial demon-stration, we benchmark several binding affinity prediction methods, including empirical scoring functions (Vina, Vinardo), a 3D convolutional neural network (Onionnet-2), and a graph neural network (AEV-PLIG). While machine learning-based models consistently outperform traditional scoring function methods, nei-ther exhibit a high correlation with ground truth affinities, highlighting the need for models specifically fine-tuned to synthetic structure distributions. This work provides a foundation for developing and evaluating next-generation structure and binding-affinity prediction models and offers insights into the structural and phys-ical underpinnings of protein-ligand interactions. The dataset can be found at https://www.sandboxaq.com/sair.

biochemistry↗

bioRxiv2023

Curiosity as an amplifier of reward-related feedback processing

Successful learning depends on various factors such as depth of processing, motivation, or curiosity about information. A strong drive to learn something or the expectation of receiving a reward can be crucial to enhance learning. However, the influence of curiosity on the processing of new information and its similarity with reward processing is not well understood. This study examined whether states of curiosity influence specific event-related-potentials (ERPs) associated with reward processing and whether these ERPs are related with later memory benefits. In an initial screening phase, participants indicated their curiosity and confidence in prior knowledge about answers to various trivia questions. In a subsequent study phase, we targeted different time windows related to reward processing during the presentation of trivia answers containing the Reward-Positivity (RewP; 250 - 350 ms), the P3 (250 - 500 ms), and the Late-Positive-Potential (LPP; 600 - 1000 ms). In a following surprise memory test, we found that participants recalled more high- than low-curiosity answers. The RewP, P3 and LPP showed greater positive mean amplitudes for high compared to low curiosity, reflecting increased reward processing. In addition, we found that the RewP and the P3 showed more positive mean amplitudes for later recalled compared to later forgotten answers, but curiosity did not modulate this encoding-related results. These findings support the view that the satisfaction of curiosity resembles reward processing, indicated by event-related-potentials.

neuroscience↗

bioRxiv2021

Goal-oriented predictive representations in the human hippocampus

Recent work in cognitive and systems neuroscience has suggested that the hippocampus might support planning, imagination, and navigation by forming "cognitive maps" that capture the structure of physical spaces, tasks, and situations. Critically, navigation involves planning within a context and disambiguating similar contexts to reach a goal. We examined hippocampal activity patterns in humans during a goal-directed navigation task to examine how contextual and goal information are incorporated in the construction and execution of navigational plans. Results demonstrate that, during planning, the hippocampus carries a context-specific representation of a future goal. Importantly, this effect could not be explained by stimulus or spatial information alone. During navigation, we observed reinstatement of activity patterns in the hippocampus ahead of participants required actions, which was strongest for behaviorally relevant points in the sequence. These results suggest that, rather than simply representing overlapping associations, hippocampal activity patterns are powerfully shaped by context and goals.

neuroscience↗