Artificial intelligence in cancer target identification and drug discovery

Signal Transduction and Targeted Therapy, Journal Year: 2022, Volume and Issue: 7(1)

Published: May 10, 2022

Artificial intelligence is an advanced method to identify novel anticancer targets and discover drugs from biology networks because the can effectively preserve quantify interaction between components of cell systems underlying human diseases such as cancer. Here, we review discuss how employ artificial approaches drugs. First, describe scope analysis for target investigations. Second, basic principles theory commonly used network-based machine learning-based algorithms. Finally, showcase applications in cancer identification drug discovery. Taken together, models have provided us with a quantitative framework study relationship network characteristics cancer, thereby leading potential discovery candidates.

Language: Английский

---

Integration of multiomics data with graph convolutional networks to identify new cancer genes and their associated molecular mechanisms

Nature Machine Intelligence, Journal Year: 2021, Volume and Issue: 3(6), P. 513 - 526

Published: April 12, 2021

Language: Английский

---

Deep Learning for Medical Image-Based Cancer Diagnosis

Cancers, Journal Year: 2023, Volume and Issue: 15(14), P. 3608 - 3608

Published: July 13, 2023

(1) Background: The application of deep learning technology to realize cancer diagnosis based on medical images is one the research hotspots in field artificial intelligence and computer vision. Due rapid development methods, requires very high accuracy timeliness as well inherent particularity complexity imaging. A comprehensive review relevant studies necessary help readers better understand current status ideas. (2) Methods: Five radiological images, including X-ray, ultrasound (US), computed tomography (CT), magnetic resonance imaging (MRI), positron emission (PET), histopathological are reviewed this paper. basic architecture classical pretrained models comprehensively reviewed. In particular, advanced neural networks emerging recent years, transfer learning, ensemble (EL), graph network, vision transformer (ViT), introduced. overfitting prevention methods summarized: batch normalization, dropout, weight initialization, data augmentation. image-based analysis sorted out. (3) Results: Deep has achieved great success diagnosis, showing good results image classification, reconstruction, detection, segmentation, registration, synthesis. However, lack high-quality labeled datasets limits role faces challenges rare multi-modal fusion, model explainability, generalization. (4) Conclusions: There a need for more public standard databases cancer. pre-training potential be improved, special attention should paid multimodal fusion supervised paradigm. Technologies such ViT, few-shot will bring surprises images.

Language: Английский

---

iGRLDTI: an improved graph representation learning method for predicting drug–target interactions over heterogeneous biological information network

Bioinformatics, Journal Year: 2023, Volume and Issue: 39(8)

Published: July 27, 2023

The task of predicting drug-target interactions (DTIs) plays a significant role in facilitating the development novel drug discovery. Compared with laboratory-based approaches, computational methods proposed for DTI prediction are preferred due to their high-efficiency and low-cost advantages. Recently, much attention has been attracted apply different graph neural network (GNN) models discover underlying DTIs from heterogeneous biological information (HBIN). Although GNN-based achieve better performance, they prone encounter over-smoothing simulation when learning latent representations drugs targets rich neighborhood HBIN, thereby reduce discriminative ability prediction.In this work, an improved representation method, namely iGRLDTI, is address above issue by capturing more feature space. Specifically, iGRLDTI first constructs HBIN integrating knowledge interactions. After that, it adopts node-dependent local smoothing strategy adaptively decide propagation depth each biomolecule thus significantly alleviating enhancing targets. Finally, Gradient Boosting Decision Tree classifier used predict DTIs. Experimental results demonstrate that yields performance several state-of-the-art on benchmark dataset. Besides, our case study indicates can successfully identify distinguishable features targets.Python codes dataset available at https://github.com/stevejobws/iGRLDTI/.

Language: Английский

---

Toward Explainable Artificial Intelligence for Precision Pathology

Annual Review of Pathology Mechanisms of Disease, Journal Year: 2023, Volume and Issue: 19(1), P. 541 - 570

Published: Oct. 23, 2023

The rapid development of precision medicine in recent years has started to challenge diagnostic pathology with respect its ability analyze histological images and increasingly large molecular profiling data a quantitative, integrative, standardized way. Artificial intelligence (AI) and, more precisely, deep learning technologies have recently demonstrated the potential facilitate complex analysis tasks, including clinical, histological, for disease classification; tissue biomarker quantification; clinical outcome prediction. This review provides general introduction AI describes developments focus on applications beyond. We explain limitations black-box character conventional describe solutions make machine decisions transparent so-called explainable AI. purpose is foster mutual understanding both biomedical side. To that end, addition providing an overview relevant foundations learning, we present worked-through examples better practical what can achieve how it should be done.

Language: Английский

---

CLARUS: An interactive explainable AI platform for manual counterfactuals in graph neural networks

Journal of Biomedical Informatics, Journal Year: 2024, Volume and Issue: 150, P. 104600 - 104600

Published: Jan. 30, 2024

Lack of trust in artificial intelligence (AI) models medicine is still the key blockage for use AI clinical decision support systems (CDSS). Although are already performing excellently medicine, their black-box nature entails that patient-specific decisions incomprehensible physician. Explainable (XAI) algorithms aim to "explain" a human domain expert, which input features influenced specific recommendation. However, domain, these explanations must lead some degree causal understanding by clinician. We developed CLARUS platform, aiming promote graph neural network (GNN) predictions. enables visualisation networks, as well as, relevance values genes and interactions, computed XAI methods, such GNNExplainer. This experts gain deeper insights into more importantly, expert can interactively alter based on acquired initiate re-prediction or retraining. interactivity allows us ask manual counterfactual questions analyse effects GNN prediction. present first interactive platform prototype, CLARUS, not only evaluation user-defined alterations patient networks outcome but also retraining entire after changing underlying structures. The currently hosted GWDG https://rshiny.gwdg.de/apps/clarus/.

Language: Английский

---

A review of cancer data fusion methods based on deep learning

Information Fusion, Journal Year: 2024, Volume and Issue: 108, P. 102361 - 102361

Published: March 20, 2024

Language: Английский

---

Artificial intelligence in drug development

Nature Medicine, Journal Year: 2025, Volume and Issue: 31(1), P. 45 - 59

Published: Jan. 1, 2025

Language: Английский

---

Graph machine learning for integrated multi-omics analysis

British Journal of Cancer, Journal Year: 2024, Volume and Issue: 131(2), P. 205 - 211

Published: May 10, 2024

Abstract Multi-omics experiments at bulk or single-cell resolution facilitate the discovery of hypothesis-generating biomarkers for predicting response to therapy, as well aid in uncovering mechanistic insights into cellular and microenvironmental processes. Many methods data integration have been developed identification key elements that explain predict disease risk other biological outcomes. The heterogeneous graph representation multi-omics provides an advantage discerning patterns suitable predictive/exploratory analysis, thus permitting modeling complex relationships. Graph-based approaches—including neural networks—potentially offer a reliable methodological toolset can provide tangible alternative scientists clinicians seek ideas implementation strategies integrated analysis their omics sets biomedical research. workflows continue push limits technological envelope, this perspective focused literature review research articles which machine learning is utilized analyses, with several examples demonstrate effectiveness graph-based approaches.

Language: Английский

---

Improving cancer driver gene identification using multi-task learning on graph convolutional network

Briefings in Bioinformatics, Journal Year: 2021, Volume and Issue: 23(1)

Published: Sept. 21, 2021

Cancer is thought to be caused by the accumulation of driver genetic mutations. Therefore, identifying cancer genes plays a crucial role in understanding molecular mechanism and developing precision therapies biomarkers. In this work, we propose Multi-Task learning method, called MTGCN, based on Graph Convolutional Network identify genes. First, augment gene features introducing their protein-protein interaction (PPI) network. After that, multi-task framework propagates aggregates nodes graph from input next layer learn node embedding features, simultaneously optimizing prediction task link task. Finally, use Bayesian weight learner balance two tasks automatically. The outputs MTGCN assign each probability being gene. Our method other four existing methods are applied predict drivers for pan-cancer some single types. experimental results show that our model shows outstanding performance compared with state-of-the-art terms area under Receiver Operating Characteristic (ROC) curves precision-recall curves. freely available via https://github.com/weiba/MTGCN.

Language: Английский

---