Mias-Lucquin, D. | bioRxiv Science

bioRxiv2025

StructGuy: Data leakage free prediction of functional effects of genetic variants.

The extent to which variations in protein-coding genes affect protein function has drawn the biological machine learning communitys attention to computationally model variant effect prediction tools. Multiplexed assays of variant effects (MAVE) experiments serve as a rich data source, but cannot deliver enough data for training truly large neural-net models. Therefore, zero-shot methods, for example protein language models, have increasingly gained popularity. For these methods, MAVE results serve primarily for evaluation purposes, as exemplified by the ProteinGym benchmark. In this study, we argue that the rapidly increasing amounts of MAVE data can be used to train efficient supervised methods, presenting our new tool StructGuy, based on gradient boosting trees methodology. In contrast to other supervised methods in the field, StructGuy, thanks to its dedicated training dataset and data leakage-free training process, can predict variant effects for proteins not seen during training. To evaluate this generalization ability, we constructed a dedicated benchmark and compared StructGuy with zero-shot methods from the ProteinGym leaderboard achieving a competitive performance. Further, we demonstrate that thanks to its architecture and careful feature engineering, we are able to provide fully interpretable predictions and direct explanations of the influence of mutations on protein three-dimensional structure, which favourably differs StructGuy from zero-shot tools.

bioinformatics↗

bioRxiv2024

Subtle changes at the RBD/hACE2 interface during SARS-CoV2 variant evolution: a molecular dynamics study

The SARS-CoV-2 Omicron variants present a different behavior compared to the previous variants, all particularly in respect to the Delta variant, as it seems to promote a lower morbidity although being much more contagious. In this perspective, we performed new molecular dynamics (MD) simulations of the various spike RBD/hACE2 complexes corresponding to the WT, Delta and Omicron variants (BA.1 up to BA.4/5) over 1.5 {micro}s timescale. Then, carrying out a comprehensive analysis of residue interactions within and between the two partners, allowed us to draw the profile of each variant by using complementary methods (PairInt, hydrophobic potential, contact PCA). Main results of PairInt calculations highlighted the most involved residues in electrostatic interactions that represent a strong contribution in the binding with highly stable contacts between spike RBD and hACE2 (importance of mutated residues at positions 417, 493 and 498). In addition to the swappable arginine residues (493/498), the apolar contacts made a substantial and complementary contribution in Omicron with the detection of two hydrophobic patches, one of which was correlated with energetic contribution calculations. This study brings new highlights on the global dynamics of spike RBD/hACE2 complexes resulting from the analysis of contact networks and cross-correlation matrices able to detect subtle changes at point mutations. The results of our study are also consistent with alternative approaches such as binding free energy calculations but are more informative and sensitive to transient or low-energy interactions. Nevertheless, the energetic contributions of residues at positions 501 and 505 were in good agreement with hydrophobic interactions measurements. The contact PCA networks could identify the intramolecular incidence of the S375F mutation occurring in all Omicron variants and likely conferring them an advantage in binding stability. Collectively, these data revealed the major differences observed between WT/Delta and Omicron variants at the RBD/hACE2 interface, which may explain the greater persistence of Omicron. Author SummaryThe evolution of SARS-CoV-2 was extremely rapid, leading to the global predominance of Omicron variants, despite the many mutations identified in the spike protein. Some of these were introduced to evade the immune system, but many others were located in the Receptor Binding Domain (RBD) without affecting its efficient binding to hACE2 and preserving the high infectivity of this variant. To unravel the mechanism by which this protein-protein connection remains strong or stable, it is necessary to study the different types of interactions at the atomic level and over time using molecular dynamics (MD) simulations. Indeed, in contrast to crystal or cryo-EM structures providing only a fixed image of the binding process, MD simulations have allowed to unambiguously identify the sustainability of some interactions mediated by key residues of spike RBD. This study could also highlight the interchangeable role of certain residues in compensating for a mutation, which in turn allows the virus to maintain durable binding to the host cell receptor. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=83 SRC="FIGDIR/small/628120v1_ufig1.gif" ALT="Figure 1"> View larger version (35K): org.highwire.dtl.DTLVardef@b2a6c4org.highwire.dtl.DTLVardef@e29044org.highwire.dtl.DTLVardef@6d9835org.highwire.dtl.DTLVardef@123c6f9_HPS_FORMAT_FIGEXP M_FIG Graphical abstract C_FIG

bioinformatics↗