Biology subjects

Husak, V.

Publications and source records attributed to Husak, V..

4 recordsLinked to original sources

bioRxiv2026

FoldARE, an RNA secondary structure analysis and prediction tool via generative pseudo-SHAPE modeling

RNA secondary structure prediction is limited by conformational heterogeneity and the scarcity of experimental data, as many RNAs populate ensembles of near-isoenergetic folds and SHAPE data are often unavailable. We present FoldARE (Folding and Analysis of RNA Ensembles), a two-step framework that derives pseudo-SHAPE constraints from in silico structural ensembles and uses them to guide SHAPE-aware secondary structure prediction. In the first step, an ensemble is generated and parsed nucleotide by nucleotide to estimate single-strandedness frequencies, which are converted into a pseudo-SHAPE reactivity profile using a weight-and-threshold scheme. In the second step, this profile is provided as a constraint to a SHAPE-compatible folding algorithm to improve the final prediction. We systematically evaluated all combinations of four ensemble-capable predictors, ViennaRNA, RNAstructure, LinearFold and EternaFold. After parameter optimization on a structurally diverse 25-RNA training set and validation using multiple scoring schemes, the best configuration combined EternaFold as ensembler and RNAstructure as predictor. Across external benchmark datasets (RNAstrand, ArchiveII and bpRNA) and the experimentally derived eFold dataset, FoldARE achieved the highest accuracy. Beyond prediction, FoldARE provides modules for ensemble-focused comparative analysis, including pairwise and multi-tool consensus assessment, per-nucleotide variability metrics, and interactive visualizations. Notably, it also supports the evaluation of m6A modification effects on structural ensembles. FoldARE is freely available on GitHub (https://github.com/TebaldiLab/FoldARE) and as a web accessible version (https://rdds.it/foldare/)

bioinformatics↗

bioRxiv2026

Critical Assessment of ML models for ADMET Prediction in TDC leaderboards

In this work we performed a critical assessment of the benchmarking procedures used in Therapeutics Data Commons (TDC) ADMET leaderboards, focusing on reproducibility, robustness against data leakage, and signs of test-set overfitting across all 22 TDC ADMET endpoints. For each endpoint, the top 3 leaderboard models were screened with a unified protocol: execution environment reproducibility check, data leakage assessment, verification of hyperparameter optimisation practices, and final re-evaluation of results and TDC ranking. Only 3 methods (CaliciBoost, MapLight, MapLight+GNN) passed all checks and showed overall reproducible performance, whereas most of top-ranked models exhibited unavailable code, non-reproducible execution environments, runtime incompatibilities, or various methodological flaws. In particular, we identified direct or indirect data leakages in MiniMol, GradientBoost and XGBoost models. We also used our in-house models based on the Mol2Vec architecture to investigate the consequences of deliberately overfitting on the TDC test set. It is shown that deliberate or accidental tuning on the public test set may lead to significant inflation of the model metrics and leaderboard position. Our results emphasize the urgent need for better public ADMET benchmarks with the hidden test sets, strict dataset versioning and model submission with standardized inference environments.

biophysics↗

bioRxiv2025

Improving ADMET prediction with descriptor augmentation of Mol2Vec embeddings

The accurate prediction of ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) properties is crucial for early-stage drug development, enabling the reduction of late-stage attrition and guiding compound prioritization. In recent years, machine learning models have emerged as powerful tools for ADMET prediction, leveraging diverse molecular representations ranging from handcrafted descriptors to graph neural networks and language model embeddings. Despite these advances, balancing predictive performance with computational efficiency remains a key challenge, particularly for high-throughput screening scenarios. Among unsupervised embedding methods, Mol2Vec has shown promise by capturing chemical substructure context analogously to word embeddings in natural language processing. However, its performance on comprehensive ADMET benchmarks has not been systematically assessed. In this work, we reimplement Mol2Vec with an expanded training corpus and higher embedding dimensionality, and evaluate its utility across 16 ADMET prediction tasks from the Therapeutics Data Commons (TDC). We show that while Mol2Vec embeddings alone are competitive, combining them with classical molecular descriptors and applying feature selection significantly improves performance. Our final MLP models with enhanced Mol2Vec embeddings achieved top-1 results in 10 of 16 benchmarks, outperforming all previously reported models on the TDC leaderboard in this regard, demonstrating that descriptor-enriched representations, paired even with relatively simple MLPs, can rival or exceed the performance of more complex models.

bioinformatics↗

bioRxiv2025

Sampling and ranking of protein conformations using machine learning techniques do not improve quality of rigid protein-protein docking

Rigid docking remains the most popular method of predicting protein-protein interactions in cases when experimental 3D structures of the complexes are not available. The docking often relies on known unbound (Apo) protein structures, which may differ significantly from their bound (Holo) forms. Modern machine learning (ML) based conformational sampling techniques allow generating ensembles of functionally relevant protein structures, which may be closer to their Holo forms and thus could improve the outcomes of the classical rigid protein-protein docking. Here, we sampled conformations of the protein subunits in 30 complexes from the novel PINDER dataset with two state-of-the-art ML-based techniques and evaluated their docking performance using several physics-based, data-based, and ML-based scoring functions. We showed that such conformational sampling rarely produces structures that are closer to the Holo conformations than the corresponding Apo ones. Moreover, even when such conformations are generated, none of the tested scoring functions were able to prioritize and rank them correctly. Our work highlights critical limitations in the current ML-enhanced rigid protein-protein docking workflows and emphasizes the need for new approaches that can better utilize the potential of modern techniques for conformational generation and scoring.

biophysics↗