Existing approaches to predicting RNA secondary structures depend on how the secondary structure is decomposed into substructures, that is, the architecture, to define their parameter space. However, architecture dependency has not been sufficiently investigated, especially for pseudoknotted secondary structures. In this study, we propose a novel algorithm for directly inferring base-pairing probabilities with neural networks that do not depend on the architecture of RNA secondary structures, and then implement this approach using two maximum expected accuracy (MEA)-based decoding algorithms: Nussinov-style decoding for pseudoknot-free structures and IPknot-style decoding for pseudoknotted structures. To train the neural networks connected to each base pair, we adopt a max-margin framework, called structured support vector machines (SSVM), as the output layer. Our benchmarks for predicting RNA secondary structures with and without pseudoknots show that our algorithm outperforms existing methods in prediction accuracy.
RNA secondary structure prediction is one of the key technologies for revealing the essential roles of functional non-coding RNAs. Although machine learning-based rich-parametrized models have achieved extremely high performance in terms of prediction accuracy, the risk of overfitting for such models has been reported. In this work, we propose a new algorithm for predicting RNA secondary structures that uses deep learning with thermodynamic integration, thereby enabling robust predictions. Similar to our previous work, the folding scores, which are computed by a deep neural network, are integrated with traditional thermodynamic parameters to enable robust predictions. We also propose thermodynamic regularization for training our model without overfitting it to the training data. Our algorithm (MXfold2) achieved the most robust and accurate predictions in computational experiments designed for newly discovered non-coding RNAs, with significant 2–10 % improvements over our previous algorithm (MXfold) and standard algorithms for predicting RNA secondary structures in terms of F -value.
Motivation: Existing approaches for predicting RNA secondary structures depend on howto decompose a secondary structure into substructures, so-called the architecture , to define their parameter space. However, the architecture has not been sufficiently investigated especially for pseudoknotted secondary structures. Results: In this paper, we propose a novel algorithm to directly infer base-pairing probabilities with neural networks that does not depend on the architecture of RNA secondary structures, followed by performing the maximum expected accuracy (MEA) based decoding algorithms; Nussinov-style decoding for pseudoknot-free structures, and IPknot-style decoding for pseudoknotted structures. To train the neural networks connected to each base-pair, we adopt a max-margin framework, called structured support vector machines (SSVM), as the output layer. Our benchmarks for predicting RNA secondary structures with and without pseudoknots show that our algorithm achieves the best prediction accuracy compared with existing methods. Availability: The source code is available at https://github.com/keio-bioinformatics/neuralfold/ . Contact: satoken@bio.keio.ac.jp
Motivation: A popular approach for predicting RNA secondary structure is the thermodynamic nearest neighbor model that finds a thermodynamically most stable secondary structure with the minimum free energy (MFE). For further improvement, an alternative approach that is based on machine learning techniques has been developed. The machine learning based approach can employ a fine-grained model that includes much richer feature representations with the ability to fit the training data. Although a machine learning based fine-grained model achieved extremely high performance in prediction accuracy, a possibility of the risk of overfitting for such model has been reported. Results: In this paper, we propose a novel algorithm for RNA secondary structure prediction that integrates the thermodynamic approach and the machine learning based weighted approach. Ourfine-grained model combines the experimentally determined thermodynamic parameters with a large number of scoring parameters for detailed contexts of features that are trained by the structured support vector machine (SSVM) with the ℓ1 regularization to avoid overfitting. Our benchmark shows that our algorithm achieves the best prediction accuracy compared with existing methods, and heavy overfitting cannot be observed. Availability: The implementation of our algorithm is available at https://github.com/keio-bioinformatics/mxfold . Contact: satoken@bio.keio.ac.jp
Motivation: Computational techniques for drug-disease prediction are essential in enhancing drug discovery and repositioning. While many methods utilize multimodal networks from various biolog-ical databases, few integrate comprehensive multi-omics data, including transcriptomes, proteomes, and metabolomes. We introduce STRGNN, a novel graph deep learning approach that predicts drug-disease relationships using extensive multimodal networks comprising proteins, RNAs, metab-olites, and compounds. We have constructed a detailed dataset incorporating multi-omics data and developed a learning algorithm with topological regularization. This algorithm selectively leverages informative modalities while filtering out redundancies. Results: STRGNN demonstrates superior accuracy compared to existing methods and has identified several novel drug effects, corroborating existing literature. STRGNN emerges as a powerful tool for drug prediction and discovery. The source code for STRGNN, along with the dataset for perfor-mance evaluation, is available at https://github.com/yuto-ohnuki/STRGNN.git.
Molecular generation models, especially chemical language model (CLM) utilizing SMILES, a string representation of compounds, face limitations in handling large and complex compounds while maintaining structural accuracy. To address these challenges, we propose FRATTVAE, a Transformer-based variational autoencoder that treats molecules as tree structures with fragments as nodes. FRATTVAE employs several innovative deep learning techniques, including ECFP (Extended Connectivity Fingerprints) based token embeddings and the Transformer’s self-attention mechanism, FRATTVAE efficiently handles large-scale compounds, improving both computational speed and generation accuracy. Evaluations across benchmark datasets, ranging from small molecules to natural compounds, demonstrate that FRATTVAE consistently outperforms existing models, achieving superior reconstruction accuracy and generation quality. Additionally, in molecular optimization tasks, FRATTVAE generated stable, high-quality molecules with desired properties, avoiding structural alerts. These results highlight FRATTVAE as a robust and versatile solution for molecular generation and optimization, making it well-suited for a variety of applications in cheminformatics and drug discovery.