TransGAT-DTA: A multi-task framework for drug-target affinity prediction and conditional molecule generation.

Identifying novel drugs that can interact with target proteins is a highly challenging, time-consuming, and costly task in drug discovery and development. Numerous machine learning-based models have recently been utilized to accelerate the drug discovery process. However, these existing methods are primarily uni-tasking, either designed to predict drug-target interaction (DTI) or generate new drugs. Through the lens of pharmacological research, these tasks are intrinsically interconnected and play a critical role in effective drug development. Therefore, the learning models must be utilized in such a manner to learn the structural properties of drug molecules, the conformational dynamics of proteins, and the bioactivity between drugs and targets. To this end, this paper develops a novel multitask learning framework that can predict drug-target binding affinities and simultaneously generate new target-aware drug variants, using common features for both tasks. In addition, we developed the FetterGrad algorithm to address the optimization challenges associated with multitask learning particularly those caused by gradient conflicts between distinct tasks. Comprehensive experiments on three real-world datasets demonstrate that the proposed model provides an effective mechanism for predicting drug-target binding affinities and generating novel drugs, thus greatly facilitating the drug discovery process.

Identifying drug-target binding affinity (DTA) plays a critical role in early-stage drug discovery. Despite the availability of various existing methods, there are still two limitations. Firstly, sequence-based methods often extract features from fixed length protein sequences, requiring truncation or padding, which can result in information loss or the introduction of unwanted noise. Secondly, structure-based methods prioritize extracting topological information but struggle to effectively capture sequence features. To address these challenges, we propose a novel deep learning model named GraphkmerDTA, which integrates Kmer features with structural topology. Specifically, GraphkmerDTA utilizes graph neural networks to extract topological features from both molecules and proteins, while fully connected networks learn local sequence patterns from the Kmer features of proteins. Experimental results indicate that GraphkmerDTA outperforms existing methods on benchmark datasets. Furthermore, a case study on lung cancer demonstrates the effectiveness of GraphkmerDTA, as it successfully identifies seven known EGFR inhibitors from a screening library of over two thousand compounds. To further assess the practical utility of GraphkmerDTA, we integrated it with network pharmacology to investigate the mechanisms underlying the therapeutic effects of Lonicera japonica flower in treating Alzheimer's disease. Through this interdisciplinary approach, three potential compounds were identified and subsequently validated through molecular docking studies. In conclusion, we present not only a novel AI model for the DTA task but also demonstrate its practical application in drug discovery by integrating modern AI approaches with traditional drug discovery methodologies.

In drug development, drug-target affinity (DTA) prediction is a key indicator for assessing the drug’s efficacy and safety. Despite significant progress in deep learning-based affinity prediction approaches in recent years, there are still limitations in capturing the complex interactions between drugs and target receptors. To address this issue, a dynamic heterogeneous graph prediction model, DynHeter-DTA, is proposed in this paper, which fully leverages the complex relationships between drug–drug, protein–protein, and drug–protein interactions, allowing the model to adaptively learn the optimal graph structures. Specifically, (1) in the data processing layer, to better utilize the similarities and interactions between drugs and proteins, the model dynamically adjusts the connection strengths between drug–drug, protein–protein, and drug–protein pairs, constructing a variable heterogeneous graph structure, which significantly improves the model’s expressive power and generalization performance; (2) in the model design layer, considering that the quantity of protein nodes significantly exceeds that of drug nodes, an approach leveraging Graph Isomorphism Networks (GIN) and Self-Attention Graph Pooling (SAGPooling) is proposed to enhance prediction efficiency and accuracy. Comprehensive experiments on the Davis, KIBA, and Human public datasets demonstrate that DynHeter-DTA exceeds the performance of previous models in drug-target interaction forecasting, providing an innovative solution for drug-target affinity prediction.

Earthquake-induced building damage can be directly devastating, leading to a large population loss and massive property damage with the high rate of globalization and urbanization. In case an earthquake takes place, the location and the scale of collapsed buildings are vital for rescuers to take effective aid measures immediately to reduce casualties and economic loss. The development in earth observation using high spatial resolution images makes it possible to recognize damaged and intact buildings. However, the methods proposed are most limited to specific cases with a limited number of buildings in pure background objects. In this paper, a multi-task deep learning framework is proposed to map damaged and intact buildings from large-scale very high spatial resolution images. The images used in our study have complicated background objects, which share similar spectral and textual characteristics with buildings. Therefore, we built a main task model to detect buildings in good shape and damaged conditions, and an extra task model to supplement the main task by semantically segmenting the input image into multiple ground objects in the multi-task framework. It is an end-to-end framework, comprising residual network and pyramid pooling module, aiming to extract multi-scale contextual features. The proposed framework is trained and evaluated in different parts of western Haiti, which gets affected by the earthquake in 2010. Besides, the results demonstrate that multi-task deep learning framework is encouraging to map earthquake-induced building damage in complicated background.

The discovery of novel drug targets is a significant challenge in drug development. Although the human genome comprises approximately 30,000 genes, proteins encoded by fewer than 400 are used as drug targets in the treatment of diseases. Therefore, novel drug targets are extremely valuable as the source for first in class drugs. On the other hand, many of the currently known drug targets are functionally pleiotropic and involved in multiple pathologies. Several of them are exploited for treating multiple diseases, which highlights the need for methods to reliably reposition drug targets to new indications. Network-based methods have been successfully applied to prioritize novel disease-associated genes. In recent years, several such algorithms have been developed, some focusing on local network properties only, and others taking the complete network topology into account. Common to all approaches is the understanding that novel disease-associated candidates are in close overall proximity to known disease genes. However, the relevance of these methods to the prediction of novel drug targets has not yet been assessed. Here, we present a network-based approach for the prediction of drug targets for a given disease. The method allows both repositioning drug targets known for other diseases to the given disease and the prediction of unexploited drug targets which are not used for treatment of any disease. Our approach takes as input a disease gene expression signature and a high-quality interaction network and outputs a prioritized list of drug targets. We demonstrate the high performance of our method and highlight the usefulness of the predictions in three case studies. We present novel drug targets for scleroderma and different types of cancer with their underlying biological processes. Furthermore, we demonstrate the ability of our method to identify non-suspected repositioning candidates using diabetes type 1 as an example.

Ricci-GraphDTA: A graph neural network integrating discrete Ricci curvature for drug-target affinity prediction.

Segment aorta and localize landmarks simultaneously on noncontrast CT using a multitask learning framework for patients without severe vascular disease

Constructing accurate gene regulatory network s (GRNs), which reflect the dynamic governing process between genes, is critical to understanding the diverse cellular process and unveiling the complexities in biological systems. With the development of computer sciences, computational-based approaches have been applied to the GRNs inference task. However, current methodologies face challenges in effectively utilizing existing topological information and prior knowledge of gene regulatory relationships, hindering the comprehensive understanding and accurate reconstruction of GRNs. In response, we propose a novel graph neural network (GNN)-based Multi-Task Learning framework for GRN reconstruction, namely MTLGRN. Specifically, we first encode the gene promoter sequences and the gene biological features and concatenate the corresponding feature representations. Then, we construct a multi-task learning framework including GRN reconstruction, Gene knockout predict, and Gene expression matrix reconstruction. With joint training, MTLGRN can optimize the gene latent representations by integrating gene knockout information, promoter characteristics, and other biological attributes. Extensive experimental results demonstrate superior performance compared with state-of-the-art baselines on the GRN reconstruction task, efficiently leveraging biological knowledge and comprehensively understanding the gene regulatory relationships. MTLGRN also pioneered attempts to simulate gene knockouts on bulk data by incorporating gene knockout information.

To develop an ensemble multi-task deep learning (DL) framework for automatic and simultaneous detection, segmentation, and classification of primary bone tumors (PBTs) and bone infections based on multi-parametric MRI from multi-center. This retrospective study divided 749 patients with PBTs or bone infections from two hospitals into a training set (N = 557), an internal validation set (N = 139), and an external validation set (N = 53). The ensemble framework was constructed using T1-weighted image (T1WI), T2-weighted image (T2WI), and clinical characteristics for binary (PBTs/bone infections) and three-category (benign/intermediate/malignant PBTs) classification. The detection and segmentation performances were evaluated using Intersection over Union (IoU) and Dice score. The classification performance was evaluated using the receiver operating characteristic (ROC) curve and compared with radiologist interpretations. On the external validation set, the single T1WI-based and T2WI-based multi-task models obtained IoUs of 0.71 ± 0.25/0.65 ± 0.30 for detection and Dice scores of 0.75 ± 0.26/0.70 ± 0.33 for segmentation. The framework achieved AUCs of 0.959 (95%CI, 0.955-1.000)/0.900 (95%CI, 0.773-0.100) and accuracies of 90.6% (95%CI, 79.7-95.9%)/78.3% (95%CI, 58.1-90.3%) for the binary/three-category classification. Meanwhile, for the three-category classification, the performance of the framework was superior to that of three junior radiologists (accuracy: 65.2%, 69.6%, and 69.6%, respectively) and comparable to that of two senior radiologists (accuracy: 78.3% and 78.3%). The MRI-based ensemble multi-task framework shows promising performance in automatically and simultaneously detecting, segmenting, and classifying PBTs and bone infections, which was preferable to junior radiologists. Compared with junior radiologists, the ensemble multi-task deep learning framework effectively improves differential diagnosis for patients with primary bone tumors or bone infections. This finding may help physicians make treatment decisions and enable timely treatment of patients. • The ensemble framework fusing multi-parametric MRI and clinical characteristics effectively improves the classification ability of single-modality models. • The ensemble multi-task deep learning framework performed well in detecting, segmenting, and classifying primary bone tumors and bone infections. • The ensemble framework achieves an optimal classification performance superior to junior radiologists' interpretations, assisting the clinical differential diagnosis of primary bone tumors and bone infections.

BackgroundThe rapid development of artificial intelligence has permeated many fields, with its application in drug discovery becoming increasingly mature. Machine learning, particularly deep learning, has significantly improved the efficiency of drug discovery. In the core task of predicting drug–target affinity (DTA), deep learning enhances predictive performance by automatically extracting complex features from compounds and proteins.MethodsTraditional approaches often rely heavily on sequence and two-dimensional structural information, overlooking critical three-dimensional and physicochemical properties. To address this, we propose a novel model—Cross Attention Fusion based on Information Enhancement for Drug–Target Affinity Prediction (CAFIE-DTA)—which incorporates protein 3D curvature and electrostatic potential information. The model approximates protein surface curvature using Delaunay triangulation, calculates total electrostatic potential via Adaptive Poisson-Boltzmann Solver (APBS) software, and employs cross multi-head attention to fuse physicochemical and sequence information of proteins. Simultaneously, it integrates graph-based and physicochemical features of compounds using the same attention mechanism. The resulting protein and compound vectors are concatenated for affinity prediction.ResultsCross-validation and comparative evaluations on the benchmark Davis and KIBA datasets demonstrate that CAFIE-DTA outperforms existing methods. On the Davis dataset, it achieved improvements of 0.003 in confidence interval (CI) and 0.022 in R2. On the KIBA dataset, it improved MSE by 0.008, CI by 0.005, and R2 by 0.017. Compared to traditional models relying on 2D structures and sequence data, CAFIE-DTA shows superior performance in DTA prediction. The source code is available at: https://github.com/NTU-MedAI/CAFIE-DTA.

The COVID-19 pandemic caused by (SARS-CoV-2) a threat, leading to numerous deaths and socioeconomic disruptions. Spurred intense research efforts effective treatment. Our study provides a brief overview COVID-19. Urgent need for effective treatments has prompted extensive research to identify potential drug targets against disease. Review shows most promising drug and their associated therapeutic approaches for combating COVID-19. Key targets include the spike protein, which facilitates viral entry into host cells, and proteases essential for viral replication. Additionally, RNA-dependent RNA polymerase (RdRp) inhibitors have been explored to inhibit viral RNA replication, highlighting their mechanisms of action, potential therapeutic benefits, and challenges in drug development. Host factors, such as the ACE2 receptor and immune response modulators, are also targeted. Combination therapies and overcoming challenges in the drug development are crucial for successful COVID-19 treatment. In this review the molecular docking study is discussed here. The future perspective of drug targets for COVID-19 encompasses a range of innovative approaches aimed at combating the virus and preparing for future outbreaks. The review also discusses the challenges faced and future directions in the field of drug target research for COVID-19. This review will provide an overview of the anticipated advancements in drug target discovery and development for COVID-19, highlighting key areas of focus and potential strategies.

The realization of high throughput sample processing has become a primary ambition in many research applications with an example being high throughput screening (HTS), which represents the first step in the drug discovery pipeline. Microfluidics is a viable platform for parallel processing of biochemical reactions to increase data production rates due to its ability to generate fluidic networks with a high number of processors over small footprints suitable for optical imaging. Single-molecule detection (SMD) is another technology which has emerged to facilitate the realization of high throughput data processing afforded by its ability to eliminate sample processing steps and generate results with high statistical accuracy. A combination of microfluidics and SMD with wide-field fluorescence detection provides the ability to monitor biochemical reactions in a high throughput format with single-molecule sensitivity. In this dissertation, the integration of these techniques was presented for HTS applications in drug discovery. An ultra-sensitive fluorescence detection system with a wide field-of-view (FoV) was constructed to transduce fluorescence signatures from single chromophores that were electrokinetically transported through a series of tightly packed fluidic channels poised on poly(methylmethacrylate), PMMA and contained within the FoV of a laser detection system. The system was used to monitor biochemical reactions at the single-molecule level in a continuous-flow format. Enhancement in sampling-throughput was demonstrated by constructing a high density fluidic network for parallel analysis of multiple biochemical assays. In another development, the ability to enhance single-molecule sensitivity in a flow-based biochemical assay was investigated using a novel cyclic olefin copolymer (COC) planar waveguide embedded in PMMA and situated orthogonal to multiple fluidic channels. This design allowed for fluorescence detection from multiple fluidic channels using evanescent excitation and a wide FoV fluorescence detection system for parallel readout. Results from these technologies were presented as well as their applications in drug discovery for increasing data production rates and quality. An approach toward monitoring the efficacy of therapeutic agents, which is important in clinical evaluation of drug potency in the drug discovery process, was also considered, by designing a microfluidic system with integrated conductivity sensor for label-free enumeration of isolated tumor cells from clinical samples.

Simple SummaryPersonalized cancer treatment strategies, including risk-adaptive chemoradiation therapy based on medical imaging, seek to improve outcomes of patients with unresectable and locally advanced non-small cell lung cancer. Refining patient risk stratification relies on outcome prediction modeling based in part on information from different imaging modalities and imaging time points during and after treatment. Using prospectively collected longitudinal data from FDG-PET, CT, and perfusion SPECT images of patients enrolled on a clinical trial, our aim was to evaluate the utility of a multitask machine learning radiomics framework for survival outcome prediction. We found that multitask learning of FDG-PET radiomics on pretreatment and mid-treatment images achieved higher survival prediction concordance compared with single-task learning of other modalities and clinical benchmark models. Our multitask learning radiomics framework can be applied to other longitudinal imaging datasets, and, once validated, can strengthen clinical decision support for personalized and adaptive treatment courses.Medical imaging provides quantitative and spatial information to evaluate treatment response in the management of patients with non-small cell lung cancer (NSCLC). High throughput extraction of radiomic features on these images can potentially phenotype tumors non-invasively and support risk stratification based on survival outcome prediction. The prognostic value of radiomics from different imaging modalities and time points prior to and during chemoradiation therapy of NSCLC, relative to conventional imaging biomarker or delta radiomics models, remains uncharacterized. We investigated the utility of multitask learning of multi-time point radiomic features, as opposed to single-task learning, for improving survival outcome prediction relative to conventional clinical imaging feature model benchmarks. Survival outcomes were prospectively collected for 45 patients with unresectable NSCLC enrolled on the FLARE-RT phase II trial of risk-adaptive chemoradiation and optional consolidation PD-L1 checkpoint blockade (NCT02773238). FDG-PET, CT, and perfusion SPECT imaging pretreatment and week 3 mid-treatment was performed and 110 IBSI-compliant pyradiomics shape-/intensity-/texture-based features from the metabolic tumor volume were extracted. Outcome modeling consisted of a fused Laplacian sparse group LASSO with component-wise gradient boosting survival regression in a multitask learning framework. Testing performance under stratified 10-fold cross-validation was evaluated for multitask learning radiomics of different imaging modalities and time points. Multitask learning models were benchmarked against conventional clinical imaging and delta radiomics models and evaluated with the concordance index (c-index) and index of prediction accuracy (IPA). FDG-PET radiomics had higher prognostic value for overall survival in test folds (c-index 0.71 [0.67, 0.75]) than CT radiomics (c-index 0.64 [0.60, 0.71]) or perfusion SPECT radiomics (c-index 0.60 [0.57, 0.63]). Multitask learning of pre-/mid-treatment FDG-PET radiomics (c-index 0.71 [0.67, 0.75]) outperformed benchmark clinical imaging (c-index 0.65 [0.59, 0.71]) and FDG-PET delta radiomics (c-index 0.52 [0.48, 0.58]) models. Similarly, the IPA for multitask learning FDG-PET radiomics (30%) was higher than clinical imaging (26%) and delta radiomics (15%) models. Radiomics models performed consistently under different voxel resampling conditions. Multitask learning radiomics for outcome modeling provides a clinical decision support platform that leverages longitudinal imaging information. This framework can reveal the relative importance of different imaging modalities and time points when designing risk-adaptive cancer treatment strategies.

Most of the existing state of the art sentiment classification techniques involve the use of pre-trained embeddings. This paper postulates a generalized representation that collates training on multiple datasets using a Multi-task learning framework. We incorporate publicly available, pre-trained embeddings with Bidirectional LSTM’s to develop the multi-task model. We validate the representations on an independent test Irony dataset that can contain several sentiments within each sample, with an arbitrary distribution. Our experiments show a significant improvement in results as compared to the available baselines for individual datasets on which independent models are trained. Results also suggest superior performance of the representations generated over Irony dataset.

Expressing the polarity of sentiment as ‘positive’ and ‘negative’ usually have limited scope compared with the intensity/degree of polarity. These two tasks (i.e. sentiment classification and sentiment intensity prediction) are closely related and may offer assistance to each other during the learning process. In this paper, we propose to leverage the relatedness of multiple tasks in a multi-task learning framework. Our multi-task model is based on convolutional-Gated Recurrent Unit (GRU) framework, which is further assisted by a diverse hand-crafted feature set. Evaluation and analysis suggest that joint-learning of the related tasks in a multi-task framework can outperform each of the individual tasks in the single-task frameworks.KeywordsMulti-task learningSingle-task learningSentiment classificationSentiment intensity prediction