AutoDock Vina is one of the most popular molecular docking tools. In latest benchmark CASF-2016 for comparative assessment scoring functions, won best power among all Modern drug discovery facing a common scenario large virtual screening hits from huge compound databases. Due to seriality characteristic algorithm, there no successful report on its parallel acceleration with GPUs. Current typically relies stack computing as well allocation resource and tasks, such VirtualFlow platform. The...

Modern drug discovery typically faces large virtual screens from huge compound databases where multiple docking tools are involved for meeting various real scenes or improving the precision of screens. Among these tools, AutoDock Vina and its numerous derivatives most popular have become standard pipeline molecular in modern discovery. Our recent Vina-GPU method realized 14-fold acceleration against on a piece NVIDIA RTX 3090 GPU one screening case. Further speedup with graphics processing...

Automated annotation of protein function is challenging. As the number sequenced genomes rapidly grows, vast majority proteins can only be annotated computationally. Nature often brings several domains together to form multi-domain and multi-functional with a possibilities, each domain may fulfill its own independently or in concerted manner neighbors. Thus, it evident that prediction problem naturally inherently Multi-Instance Multi-Label (MIML) learning tasks. Based on state-of-the-art...

Autism spectrum disorder (ASD) is a complex neurodevelopmental mainly showed atypical social interaction, communication, and restricted, repetitive patterns of behavior, interests activities. Now clinic diagnosis ASD mostly based on psychological evaluation, clinical observation medical history. All these behavioral indexes could not avoid defects such as subjectivity reporter-dependency. Therefore researchers devoted themselves to seek relatively stable biomarkers supplementary diagnostic...

The identification of RNA-binding residues in proteins is important several areas such as protein function, posttranscriptional regulation and drug design. We have developed PRBR (Prediction RNA Binding Residues), a novel method for identifying from amino acid sequences. Our combines hybrid feature with the enriched random forest (ERF) algorithm. composed predicted secondary structure information three features: evolutionary combined conservation physicochemical properties acids about...

Hashing-based image retrieval methods have become a cutting-edge topic in the information domain due to their high efficiency and low cost. In order perform efficient hash learning by simultaneously preserving semantic similarity data structures feature space, this paper presents semi-supervised metric learning-based anchor graph hashing method. Our proposed approach can be divided into three parts. First, we exploit transformation matrix construct anchor-based of training set. Second,...

Precise assessment of ligand bioactivities (including IC50, EC50, Ki, Kd, etc.) is essential for virtual screening and lead compound identification. However, not all ligands have experimentally determined activities. In particular, many G protein-coupled receptors (GPCRs), which are the largest integral membrane protein family represent targets nearly 40% drugs on market, lack published experimental data about interactions. Computational methods with ability to accurately predict bioactivity...

Simulated Annealing (SA) algorithm is not effective with large optimization problems for its slow convergence. Hence, several parallel (pSA) methods have been proposed, where the increase of searching threads can boost speed Although satisfactory solutions be obtained by these methods, there no rigorous mathematical analyses on their effectiveness. Thus, this article introduces a probabilistic model, which theorem about effectiveness multiple initial states SA (MISPSA) has proven. The also...

This study presents SDL-IUAs, a novel deep learning framework that employs self-attention mechanism to predict pregnancy outcomes in patients with intrauterine adhesions (IUAs) following hysteroscopic adhesiolysis. By integrating random forest-based feature importance analysis self-attention, the model achieves enhanced predictive accuracy and interpretability. The experiments were conducted on clinical dataset of 121 163 features, where 16.05% missing values effectively handled using matrix...

Virtual screening has been widely used to identify potential hit candidates that can bind the target protein in drug discovery. Contemporary methods typically rely on oversimplified scoring functions, frequently yielding one-digit rates (or even zero) among top-ranked candidates. The substantial cost of laboratory validation further constrains exploration candidate molecules. We find test-time prediction refinement is almost blank this area, which means bioactivity feedback wet-lab...

The recognition of microRNA (miRNA)-binding residues in proteins is helpful to understand how miRNAs silence their target genes. It difficult use existing computational method predict miRNA-binding due the lack training examples. To address this issue, unlabeled data may be exploited help construct a model. Semisupervised learning deals with methods for exploiting addition labeled automatically improve performance, where no human intervention assumed. In addition, almost always contain much...

AutoDock Vina and its derivatives have established themselves as a prevailing pipeline for virtual screening in contemporary drug discovery. Our Vina-GPU method leverages the parallel computing power of GPUs to accelerate Vina, 2.0 further enhances speed derivatives. Given prevalence large screens modern discovery, improvement accuracy has become longstanding challenge. In this study, we propose 2.1, aimed at enhancing docking precision through integration novel algorithms facilitate...

DNA-binding proteins are fundamentally important in understanding cellular processes. Thus, the identification of has particularly practical application various fields, such as drug design. We have proposed a novel approach method for predicting using only sequence information. The prediction model developed this study is constructed by support vector machine-sequential minimal optimization (SVM-SMO) algorithm conjunction with hybrid feature. feature incorporating evolutionary information...

Molecular docking (MD) is one of the core steps in expensive and time-consuming process drug design, which basically an optimization problem based on scoring functions. AutoDock series MD software widely accepted by academia industry, among Vina (Vina) latest most popular version due to its accuracy relatively high speed. However, contrast prior version, i.e., AutoDock4, hardware acceleration approaches are rarely reported. In this article, we propose Vina-field-programmable gate array...

Accurate prediction and interpretation of ligand bioactivities are essential for virtual screening drug discovery. Unfortunately, many important targets lack experimental data about the bioactivities; this is particularly true G protein-coupled receptors (GPCRs), which account a third drugs currently on market. Computational approaches with potential precise assessment determination key substructural features determine needed to address issue.A new method, SED, was proposed predict recognize...

Alzheimer’s disease (AD) is a neurodegenerative brain in the elderly. Identifying patients with mild cognitive impairment (MCI) who are more likely to progress AD key step prevention. Recent studies have shown that heterogeneous disease. In this study, we propose subtyping-based prediction strategy predict conversion from MCI three years according patient subtypes. Structural magnetic resonance imaging (sMRI) data and multi-omics data, including genotype gene expression profiling derived...

Most of the news headline generation models that use sequence-to-sequence model or recurrent network have two shortcomings: lack parallel ability and easily repeated words. It is difficult to select important words in reproduce these expressions, resulting inaccurately summarizes news. In this work, we propose a TD-NHG model, which stands for based on an improved decoder from transformer. The uses masked multi-head self-attention learn feature information different representation subspaces...

Mild Cognitive Impairment (MCI) is a preclinical stage of Alzheimer's Disease (AD) and clinical heterogeneity. The classification MCI crucial for the early diagnosis treatment AD. In this study, we investigated potential using both labeled unlabeled samples from Neuroimaging Initiative (ADNI) cohort to classify through multimodal co-training method. We utilized structural magnetic resonance imaging (sMRI) data genotype 364 including 228 136 ADNI-1 cohort. First, selected quantitative trait...

Abstract Ligand molecules naturally constitute a graph structure. Recently, many excellent deep learning (DGL) methods have been proposed and used to model ligand bioactivities, which is critical for the virtual screening of drug hits from compound databases in interest. However, pharmacists can find that these well-trained DGL models usually are hard achieve satisfying performance real scenarios candidates. The main challenges involve datasets training were small-sized biased, inner active...

Accurate modeling of compound bioactivities is essential for the virtual screening drug leads. In real-world scenarios, pharmacists tend to choose from top-k hit compounds ranked by predicted a large database with interest continue wet experiments discovery. Significant improvement precision top hits in ligand-based leads more valuable than conventional schemes accurately predicting all database. Here, we proposed new method, RealVS, significantly improve hits' and learn interpretable key...

Simulated Annealing (SA) algorithm is not effective with large optimization problems for its slow convergence. Hence, several parallel (pSA) methods have been proposed, where the increase of searching threads can boost speed Although these obtain satisfactory solutions to problems, there no rigorous mathematical analyses on their effectiveness. Thus, this paper introduces a probabilistic model, which theorem about effectiveness multiple initial states SA (MISPSA) has proven. The also...

G protein-coupled receptors (GPCRs) are one of the most important drug targets, accounting for ∼34% drugs on market. For discovery, accurate modeling and explanation bioactivities ligands is critical screening optimization hit compounds. Homologous GPCRs more likely to interact with chemically similar ligands, they tend share common binding modes ligand molecules. The inclusion homologous in learning potentially enhances accuracy interpretability models due utilizing increased training...

In multi-instance multi-label learning (MIML) problems, predicting the labels of unseen bags becomes difficult when their instances are not provided directly. Therefore, it is necessary to exploit label correlations enhance accuracy MIML classification. This paper presents metric learning-based MIML-kNN (MI(ML)2kNN) method, which composed three parts. First, Laplacian matrix can be learned obtain by minimizing manifold regularizer. Then, based on correlations, a novel objective function for...