Dimitri Gilis

Food allergies represent a significant and growing global challenge. However, researchers and risk assessors still lack a dedicated and comprehensive resource that centralizes food allergen information. As a result, obtaining the necessary structural, biochemical, and immunological data often requires time-consuming searches across multiple heterogeneous databases, hindering efficient analysis and slowing scientific progress. The Food Allergen Database (FAD) merges biochemical, structural, immunological, and functional properties of all known food allergens and isoallergens (1168 entries). It is an essential resource that facilitates the extraction of all information about food allergens.

Searching for molecules optimizing certain properties remains a key challenge due to the vastness of the chemical space, its discrete nature, and the limited availability of bioactivity data. One way to address these issues is to build a mapping from the chemical space to a continuous latent embedding space where efficient exploration and smooth interpolations become possible. Existing methods suffer from several limitations: failure to accurately reconstruct molecules, poorly structured latent space where whole regions cannot be decoded to valid molecular graphs or where distances fail to reflect chemical similarities, not to mention the unavailability of ready-to-use code or models trained on sufficiently large datasets, limiting their practical application. In this work, we provide a large-scale pre-trained Variational Autoencoder based on the Transformer architecture to convert small organic molecules to continuous fixed-size embeddings in a chemistry-aware structured latent space and back. With a 97% reconstruction rate and a 100% validity rate, the model provides a reliable mapping and is able to generate a high diversity of novel molecules. We introduce a training objective with a novel loss term explicitly enforcing embedding distances to reflect Tanimoto similarities between molecular fingerprints, leading to a smooth and well-structured latent space enabling seamless exploration and interpolation. We show that our model achieves solid performances on molecular optimization tasks compared to other generative models. We release an open-source, user-friendly implementation as a pip package and at github.com/3BioCompBio/chembed, along with pre-trained models, that can be readily used or fine-tuned for downstream drug design applications. Scientific contribution: We introduce a large-scale SELFIES Transformer VAE with a novel loss aligning latent space distances with chemical similarity. This yields a highly structured, chemistry-aware latent space, alongside high reconstruction accuracy. Our open-source release with pre-trained weights makes the model directly usable and adaptable for molecular design applications.

Community‐wide blind prediction experiments such as CAPRI and CASP provide an objective measure of the current state of predictive methodology. Here we describe a community‐wide assessment of methods to predict the effects of mutations on protein–protein interactions. Twenty‐two groups predicted the effects of comprehensive saturation mutagenesis for two designed influenza hemagglutinin binders and the results were compared with experimental yeast display enrichment data obtained using deep sequencing. The most successful methods explicitly considered the effects of mutation on monomer stability in addition to binding affinity, carried out explicit side‐chain sampling and backbone relaxation, evaluated packing, electrostatic, and solvation effects, and correctly identified around a third of the beneficial mutations. Much room for improvement remains for even the best techniques, and large‐scale fitness landscapes should continue to provide an excellent test bed for continued evaluation of both existing and new prediction methodologies. Proteins 2013; 81:1980–1987.

The ability of proteins to establish highly selective interactions with a variety of (macro)molecular partners is a crucial prerequisite to the realization of their biological functions. The availability of computational tools to evaluate the impact of mutations on protein–protein binding can therefore be valuable in a wide range of industrial and biomedical applications, and help rationalize the consequences of non-synonymous single-nucleotide polymorphisms. BeAtMuSiC (http://babylone.ulb.ac.be/beatmusic) is a coarse-grained predictor of the changes in binding free energy induced by point mutations. It relies on a set of statistical potentials derived from known protein structures, and combines the effect of the mutation on the strength of the interactions at the interface, and on the overall stability of the complex. The BeAtMuSiC server requires as input the structure of the protein–protein complex, and gives the possibility to assess rapidly all possible mutations in a protein chain or at the interface, with predictive performances that are in line with the best current methodologies.

AQPs (aquaporins) are conserved in all kingdoms of life and facilitate the rapid diffusion of water and/or other small solutes across cell membranes. Among the different plant AQPs, PIPs (plasma membrane intrinsic proteins), which fall into two phylogenetic groups, PIP1 and PIP2, play key roles in plant water transport processes. PIPs form tetramers in which each monomer acts as a functional channel. The intermolecular interactions that stabilize PIP oligomer complexes and are responsible for the resistance of PIP dimers to denaturating conditions are not well characterized. In the present study, we identified a highly conserved cysteine residue in loop A of PIP1 and PIP2 proteins and demonstrated by mutagenesis that it is involved in the formation of a disulfide bond between two monomers. Although this cysteine seems not to be involved in regulation of trafficking to the plasma membrane, activity, substrate selectivity or oxidative gating of ZmPIP1s (Zm is Zea mays), ZmPIP2s and hetero-oligomers, it increases oligomer stability under denaturating conditions. In addition, when PIP1 and PIP2 are co-expressed, the loop A cysteine of ZmPIP1;2, but not that of ZmPIP2;5, is involved in the mercury sensitivity of the channels.

Allergy represents an increasing thread to public health in both developed and emerging countries and the dust mites Dermatophagoides pteronyssinus (Der p), Blomia tropicalis (Blo t), Dermatophagoides farinae (Der f), Lepidoglyphus destructor (Lep d) and Suidasia medanensis (Sui m) strongly contribute to this problem. Their allergens are classified in several families among which families 5 and 21 which are the subject of this work. Indeed, their biological function as well as the mechanism or epitopes by which they are contributing to the allergic response remain unknown and their tridimensional structures have not been resolved experimentally except for Blo t 5 and Der p 5. Blo t 5 is a monomeric three helical bundle, whereas Der p 5 shows a three helical bundle with a kinked N-terminal helix that assembles in an entangled dimeric structure with a large hydrophobic cavity. This cavity could be involved in the binding of hydrophobic ligands, which in turn could be responsible for the shift of the immune response from tolerance to allergic inflammation. We used molecular modelling approaches to bring out if other house dust mite allergens of families 5 and 21 (Der f 5, Sui m 5, Lep d 5, Der p 21 and Der f 21) could dimerize and form a large cavity in the same way as Der p 5. Monomeric models were first performed with MODELLER using the experimental structures of Der p 5 and Blo t 5 as templates. The ClusPro server processed the selected monomers in order to assess their capacity to form dimeric structures with a positive result for Der p 5 and Der f 5 only. The other allergens (Blo t 5, Sui m 5, Lep d 5, Der p 21 and Der f 21) did not present such a propensity. Moreover, we identified mutations that should destabilize and/or prevent the formation of the Der p 5 dimeric structure. The production of these mutated proteins could help us to understand the role of the dimerization process in the allergic response induced by Der p 5, and if Der p 5 and Der f 5 behave similarly.

PoPMuSiC (http://babylone.ulb.ac.be/popmusic) is a program that has been developed to predict rapidly changes in protein thermodynamic stability upon single-site mutations. We describe in this paper the theoretical model that underlies the PoPMuSiC software, and present a few applications to various issues of biological interest. In particular, we investigate the possible use of PoPMuSiC for the prediction of changes in protein-protein binding affinity upon mutation. We also summarize previous studies where PoPMuSiC has been used to modulate the relative stability of different protein structural states, to help the identification of mutations that stabilize and solubilize the Tobacco Etch Virus protease (TEV) and to detect structural weaknesses in proteins that are subject to domain swapping.

Spinocerebellar Ataxia Type 3 (SCA3) is one of nine polyglutamine (polyQ) diseases that are all characterized by progressive neuronal dysfunction and the presence of neuronal inclusions containing aggregated polyQ protein, suggesting that protein misfolding is a key part of this disease. Ataxin‐3, the causative protein of SCA3, contains a globular, structured N‐terminal domain (the Josephin domain) and a flexible polyQ‐containing C‐terminal tail, the repeat‐length of which modulates pathogenicity. It has been suggested that the fibrillogenesis pathway of ataxin‐3 begins with a non‐polyQ‐dependent step mediated by Josephin domain interactions, followed by a polyQ‐dependent step. To test the involvement of the Josephin domain in ataxin‐3 fibrillogenesis, we have created both pathogenic and nonpathogenic length ataxin‐3 variants with a stabilized Josephin domain, and have both stabilized and destabilized the isolated Josephin domain. We show that changing the thermodynamic stability of the Josephin domain modulates ataxin‐3 fibrillogenesis. These data support the hypothesis that the first stage of ataxin‐3 fibrillogenesis is caused by interactions involving the non‐polyQ containing Josephin domain and that the thermodynamic stability of this domain is linked to the aggregation propensity of ataxin‐3.

The rational design of modified proteins with controlled stability is of extreme importance in a whole range of applications, notably in the biotechnological and environmental areas, where proteins are used for their catalytic or other functional activities. Future breakthroughs in medical research may also be expected from an improved understanding of the effect of naturally occurring disease-causing mutations on the molecular level.