Disease gene prediction with privileged information and heteroscedastic dropout

Shu, Juan; Liu, Yu; Wang, Sheng; Xi, Bowei; Ma, Jianzhu

doi:10.1093/bioinformatics/btab310

Cited by 18 publications

(7 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We compare TXGNN to a state-of-the-art GNN 18 previously validated for therapeutic use prediction [6][7][8][19][20][21] . The GNN performs poorly on realistic yet challenging systematic data splits.…”

Section: Constructed)mentioning

confidence: 99%

A foundation model for clinician-centered drug repurposing

Huang

Chandak

Wang

et al. 2023

Preprint

View full text Add to dashboard Cite

Of the several thousand diseases that affect humans, only about 500 have treatments approved by the U.S. Food and Drug Administration. Even for those with approved treatments, discovering new drugs can offer alternative options that cause fewer side effects and replace drugs that are ineffective for certain patient groups. However, identifying new therapeutic opportunities for diseases with limited treatment options remains a challenge, as existing algorithms often perform poorly. Here, we leverage recent advances in geometric deep learning and human-centered AI to introduce TxGNN, a model for identifying therapeutic opportunities for diseases with limited treatment options and minimal molecular understanding. TxGNN is a graph neural network pre-trained on a comprehensive knowledge graph of 17,080 clinically-recognized diseases and 7,957 therapeutic candidates. The model can process various therapeutic tasks, such as indication and contraindication prediction, in a unified formulation. Once trained, we show that TxGNN can perform zero-shot inference on new diseases without additional parameters or fine-tuning on ground truth labels. Evaluation of TxGNN shows significant improvements over existing methods, with up to 49.2% higher accuracy in indication tasks and 35.1% higher accuracy in contraindication tasks. TxGNN can also predict therapeutic use for new drugs developed since June 2021. To facilitate interpretation and analysis of the model's predictions by clinicians, we develop a human-AI explorer for TxGNN and evaluate its usability with medical experts. Finally, we demonstrate that TxGNN's novel predictions are consistent with off-label prescription decisions made by clinicians in a large healthcare system. These label-efficient and clinician-centered learning systems pave the way for improvements for many therapeutic tasks.

show abstract

Section: Constructed)mentioning

confidence: 99%

A foundation model for clinician-centered drug repurposing

Huang

Chandak

Wang

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…RGCN [42]: As part of RGCN, disease similarities, gene similarities, and disease-gene associations are used to construct a multi-relational network. Link prediction is used here to model the disease gene prioritization problem.…”

Section: Performance Evaluationmentioning

confidence: 99%

“…Graph convolutional networks have a high level of capability in learning graph structures and have been successfully used for a variety of problems and applications, including drug discovery [39] and molecular property prediction [40]. The majority of graph convolutional networks methods [41][42][43][44] consider the problem of candidate disease gene prioritization as a link prediction problem and use convolutional graph networks and graph embedding on integrated networks. The presented methods are limited by the lack of use of other data sources with a non-graph structure and inadequate features for each node.…”

Section: Introductionmentioning

confidence: 99%

A novel candidate disease gene prioritization method using deep graph convolutional networks and semi-supervised learning

Azadifar

Ahmadi

2022

BMC Bioinformatics

View full text Add to dashboard Cite

Background Selecting and prioritizing candidate disease genes is necessary before conducting laboratory studies as identifying disease genes from a large number of candidate genes using laboratory methods, is a very costly and time-consuming task. There are many machine learning-based gene prioritization methods. These methods differ in various aspects including the feature vectors of genes, the used datasets with different structures, and the learning model. Creating a suitable feature vector for genes and an appropriate learning model on a variety of data with different and non-Euclidean structures, including graphs, as well as the lack of negative data are very important challenges of these methods. The use of graph neural networks has recently emerged in machine learning and other related fields, and they have demonstrated superior performance for a broad range of problems. Methods In this study, a new semi-supervised learning method based on graph convolutional networks is presented using the novel constructing feature vector for each gene. In the proposed method, first, we construct three feature vectors for each gene using terms from the Gene Ontology (GO) database. Then, we train a graph convolution network on these vectors using protein–protein interaction (PPI) network data to identify disease candidate genes. Our model discovers hidden layer representations encoding in both local graph structure as well as features of nodes. This method is characterized by the simultaneous consideration of topological information of the biological network (e.g., PPI) and other sources of evidence. Finally, a validation has been done to demonstrate the efficiency of our method. Results Several experiments are performed on 16 diseases to evaluate the proposed method's performance. The experiments demonstrate that our proposed method achieves the best results, in terms of precision, the area under the ROC curve (AUCs), and F1-score values, when compared with eight state-of-the-art network and machine learning-based disease gene prioritization methods. Conclusion This study shows that the proposed semi-supervised learning method appropriately classifies and ranks candidate disease genes using a graph convolutional network and an innovative method to create three feature vectors for genes based on the molecular function, cellular component, and biological process terms from GO data.

show abstract

“…TAS2Rs are a type of chemosensory G protein-coupled receptors, with a total of 25 functional gene-encoded bitter taste receptor proteins (TAS2R1, 3,4,5,7,8,9,10,13,14,16,19,20,30,31,38,39,40,41,42,43,45,46,50,60) found in human cells [14]. These receptors are not only expressed in the oral cavity but also exist in the gastrointestinal tract and other extrabuccal tissues [19,20].…”

Section: Introductionmentioning

confidence: 99%

Comprehensive Analysis of the Function and Prognostic Value of TAS2Rs Family-Related Genes in Colon Cancer

Bi,

Zhu,

Huang

et al. 2024

IJMS

View full text Add to dashboard Cite

In the realm of colon carcinoma, significant genetic and epigenetic diversity is observed, underscoring the necessity for tailored prognostic features that can guide personalized therapeutic strategies. In this study, we explored the association between the type 2 bitter taste receptor (TAS2Rs) family-related genes and colon cancer using RNA-sequencing and clinical datasets from The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO). Our preliminary analysis identified seven TAS2Rs genes associated with survival using univariate Cox regression analysis, all of which were observed to be overexpressed in colon cancer. Subsequently, based on these seven TAS2Rs prognostic genes, two colon cancer molecular subtypes (Cluster A and Cluster B) were defined. These subtypes exhibited distinct prognostic and immune characteristics, with Cluster A characterized by low immune cell infiltration and less favorable outcomes, while Cluster B was associated with high immune cell infiltration and better prognosis. Finally, we developed a robust scoring system using a gradient boosting machine (GBM) approach, integrated with the gene-pairing method, to predict the prognosis of colon cancer patients. This machine learning model could improve our predictive accuracy for colon cancer outcomes, underscoring its value in the precision oncology framework.

show abstract

Disease gene prediction with privileged information and heteroscedastic dropout

Cited by 18 publications

References 55 publications

A foundation model for clinician-centered drug repurposing

A foundation model for clinician-centered drug repurposing

A novel candidate disease gene prioritization method using deep graph convolutional networks and semi-supervised learning

Comprehensive Analysis of the Function and Prognostic Value of TAS2Rs Family-Related Genes in Colon Cancer

Contact Info

Product

Resources

About