Reboul, E. | Search bioRxiv

bioRxiv2025

Menger_Curvature : a MDAKit implementation to decipher the dynamics, curvatures and flexibilities of polymeric backbones at the residue level

Characterizing the dynamics of the backbone of flexible polymers such as Intrinsically Disordered Regions and Proteins (IDRs and IDPs) has proven to be a significant challenge in molecular dynamics (MD) simulations due to their high conformational variability. The widely-used mobility metric Root-Mean-Squared Fluctuations (RMSF) is powerless to provide information for highly flexible systems, as defining a relevant reference structure is often not possible. We previously introduced a new flexibility metric to remedy this gap : the Local Flexibilities (LFs), derived (alongside the Local Curvatures (LCs)) from the Proteic Menger Curvatures (PMCs). Here we present a numba accelerated implementation for any polymer of the calculation of Menger Curvatures as a MDAKit from the widely-used MDAnalysis package. We perform a benchmark with the RMSF and another flexibility metric derived from Proteic Blocks (PBs), the Equivalent Number of PBs (Neq), and show that the PMCs are an order of magnitude faster to compute on a modern CPU chip. We applied all 3 flexibility metrics to a {beta}III-tubulin monomer as an example, as tubulins are known to possess the entire range of proteic elements, from -helix and {beta}-sheets to a flexible loop and a disordered C-terminal tail (CTT). RMSF, LFs and Neq all succeed in identifying the flexible loops and the CTT, although the RMSF requires a system-specific alignment to do so. Finally, we expose different applications of PMCs, LCs and LFs ranging from mechanism characterization to NMR T2 predictions. We believe that Menger curvatures will prove to be a valuable metric to study protein dynamics and polymers in general. The MDAKit package Menger_Curvature is readily available at https://github.com/EtienneReboul/menger_curvature

bioinformatics↗

bioRxiv2024

Improving the reliability of molecular string representations for generative chemistry

Generative modeling for chemistry has advanced rapidly in recent years, but this surge in popularity raises a foundational question: which molecular representation is best suited for modern machine learning models? Despite not being designed for generative tasks, SMILES remain the most commonly used string-based representation. However, while SMILES follow strict syntactic rules, grammatically correct SMILES strings do not always correspond to valid molecules. SELFIES were introduced as an alternative that addresses this limitation by ensuring that every string of SELFIES tokens represents to a valid molecule. In this study, we comprehensively evaluate the limitations of both SMILES and SELFIES as representations for generative models. We define two key criteria for robust molecular generation: viability, generated strings represent novel, unique molecules with correct valence, and fidelity, the distribution of physicochemical properties from sampled molecules resembles that of the training data. We find that approximately one-fifth of molecules generated using canonical SMILES are invalid, failing the viability criterion. In contrast, all SELFIES-generated molecules are viable, but they deviate significantly from the training distribution, indicating low fidelity. To address these limitations, we develop data augmentation procedures for both representations. While simplifying the SELFIES grammar yields only modest gains in fidelity, our stochastic augmentation method for SMILES, ClearSMILES, significantly improves both viability and fidelity. ClearSMILES simplifies syntax by reducing the vocabulary size and explicitly encoding aromaticity via Kekule SMILES, making it easier string representations for models to process. Using ClearSMILES, the rate of invalid samples decreases by an order of magnitude, from 20% to 2.2%, and fidelity to the training distribution is also moderately improved. Generative chemistry has seen rapid development recently. However, models based on string representations of molecules still rely largely on SMILES1 that have not been developed for this context and SELFIES 2 who were introduced to reduce those problems. The goal of this study is to first analyze the difficulty encountered by a small generative model when using SMILES and SELFIES. Our study found that SELFIES and canonical SMILES 3 are not fully reliable representations for a small generative model, i.e. do not ensure concurrently the viability and fidelity of samples. Viable samples represent novel, unique molecules with correct valence, while fidelity is efficient distribution learning of key physico-chemical properties. 4 In fact, 20% of the samples generated using canonical SMILES input representation do not correspond to valid molecules. In contrast, samples generated using SELFIES were all viable but where not able to reproduce as well the distribution of physico-chemical properties as SMILES. As a mitigation strategy for the previously identified problems, we have developed data augmentation procedures for both SELFIES and SMILES. Simplifying the complex syntax of SELFIES yielded only marginal improvements in string stability and overall fidelity to the training set. For SMILES, we developed a stochastic data augmentation procedure called ClearSMILES, which reduces the vocabulary size needed to represent a SMILES dataset, explicitly represents aromaticity via Kekule SMILES, 3 and reduces the effort required by deep learning models to process SMILES. ClearSMILES reduced the rate of invalid samples by an order of magnitude, from 20% to 2.2%, and improved the fidelity of samples to the training set.

bioinformatics↗

bioRxiv2024

Molecular modeling of Proteinase-Activated Receptor 1 in complex with Thrombin Receptor Activator Peptide 6.

The protease-activated receptor 1 (PAR1) and its activator thrombin re-ceptor activator peptide 6 (TRAP6) play crucial roles in various physiologi-cal and pathological processes, including hemostasis, thrombosis, and cancer progression. Although the interaction between PAR1 and TRAP6 has been heavily studied using experimental technique such as mutagenesis, structural data remains scarce due to the technical hardship of studying membrane pro-teins such as PAR1. In this study, we employed an integrative modeling approach to elucidate the structure of the PAR1-TRAP6 complex. Leveraging state-of-the-art AI-based protein modeling tools, including AlphaFold2 and ESMFOLD, we in-tegrated HADDOCK, a physics-based method to refine predictions. Overall, the predicted structures are in good agreement with the experimental data available in the literature. Our model unveiled a new T-shaped pi-stacking interaction between TRAP6s F2 and PAR1s Y360. The integrative modeling approach combining the predictions of the deep learning model with a physics-based method proves to be an interesting strat-egy for solving challenging membrane protein structures with high confidence. Our model of the PAR1-TRAP6 complex will be an interesting starting point for further investigation of the activation of PAR1 by TRAP6.

bioinformatics↗