For the risk, progression, and response to treatment of many complex diseases, it has been increasingly recognized that genetic interactions (including gene-gene and gene-environment interactions) play important roles beyond the main genetic and environmental effects. In practical genetic interaction analyses, model mis-specification and outliers/contaminations in response variables and covariates are not uncommon, and demand robust analysis methods. Compared with their nonrobust counterparts, robust genetic interaction analysis methods are significantly less popular but are gaining attention fast. In this article, we provide a comprehensive review of robust genetic interaction analysis methods, on their methodologies and applications, for both marginal and joint analysis, and for addressing model mis-specification as well as outliers/contaminations in response variables and covariates.

/pdf/robust-genetic-interaction-analysis-4zyayxu4qo.pdf

Robust genetic interaction analysis

Gene-environment interactions have important implications for elucidating the genetic basis of complex diseases beyond the joint function of multiple genetic factors and their interactions (or epistasis). In the past, G × E interactions have been mainly conducted within the framework of genetic association studies. The high dimensionality of G × E interactions, due to the complicated form of environmental effects and the presence of a large number of genetic factors including gene expressions and SNPs, has motivated the recent development of penalized variable selection methods for dissecting G × E interactions, which has been ignored in the majority of published reviews on genetic interaction studies. In this article, we first survey existing studies on both gene-environment and gene-gene interactions. Then, after a brief introduction to the variable selection methods, we review penalization and relevant variable selection methods in marginal and joint paradigms, respectively, under a variety of conceptual models. Discussions on strengths and limitations, as well as computational aspects of the variable selection methods tailored for G × E studies, have also been provided.

Gene-Environment Interaction: A Variable Selection Perspective.

Gene-environment interactions have important implications to elucidate the genetic basis of complex diseases beyond the joint function of multiple genetic factors and their interactions (or epistasis). In the past, G$\times$E interactions have been mainly conducted within the framework of genetic association studies. The high dimensionality of G$\times$E interactions, due to the complicated form of environmental effects and presence of a large number of genetic factors including gene expressions and SNPs, has motivated the recent development of penalized variable selection methods for dissecting G$\times$E interactions, which has been ignored in majority of published reviews on genetic interaction studies. In this article, we first survey existing overviews on both gene-environment and gene-gene interactions. Then, after a brief introduction on the variable selection methods, we review penalization and relevant variable selection methods in marginal and joint paradigms respectively under a variety of conceptual models. Discussions on strengths and limitations, as well as computational aspects of the variable selection methods tailored for G$\times$E studies have also been provided.

Gene-Environment Interaction: A Variable Selection Perspective

In biomedical and epidemiological studies, gene–environment (G–E) interactions have been shown to importantly contribute to the etiology and progression of many complex diseases. Most existing appr...

/pdf/robust-gene-environment-interaction-analysis-using-penalized-3kfnfqil76.pdf

Robust gene-environment interaction analysis using penalized trimmed regression.

Introduction Bitter peptides are short peptides with potential medical applications. The huge potential behind its bitter taste remains to be tapped. To better explore the value of bitter peptides in practice, we need a more effective classification method for identifying bitter peptides. Methods In this study, we developed a Random forest (RF)-based model, called Bitter-RF, using sequence information of the bitter peptide. Bitter-RF covers more comprehensive and extensive information by integrating 10 features extracted from the bitter peptides and achieves better results than the latest generation model on independent validation set. Results The proposed model can improve the accurate classification of bitter peptides (AUROC = 0.98 on independent set test) and enrich the practical application of RF method in protein classification tasks which has not been used to build a prediction model for bitter peptides. Discussion We hope the Bitter-RF could provide more conveniences to scholars for bitter peptide research.

/pdf/bitter-rf-a-random-forest-machine-model-for-recognizing-39xw0bar.pdf

Bitter-RF: A random forest machine model for recognizing bitter peptides

MOTIVATION
Identification of Drug-Target Interactions (DTIs) is an essential step in drug discovery and repositioning. DTI prediction based on biological experiments is time-consuming and expensive. In recent years, graph learning based methods have aroused widespread interest and shown certain advantages on this task, where the DTI prediction is often modeled as a binary classification problem of the nodes composed of drug and protein pairs. Nevertheless, in many real applications, labeled data are very limited and expensive to obtain. With only a few thousand labeled data, models could hardly recognize comprehensive patterns of DPP node representations, and are unable to capture enough commonsense knowledge, which is required in DTI prediction. Supervised contrastive learning gives an aligned representation of DPP node representations with the same class label. In embedding space, DPP node representations with the same label are pulled together, and those with different labels are pushed apart.


RESULTS
We propose an end-to-end supervised graph co-contrastive learning model for DTI prediction directly from heterogeneous networks. By contrasting the topology structures and semantic features of the drug-protein-pair network, as well as the new selection strategy of positive and negative samples, SGCL-DTI generates a contrastive loss to guide the model optimization in a supervised manner. Comprehensive experiments on three public datasets demonstrate that our model outperforms the SOTA methods significantly on the task of DTI prediction, especially in the case of cold start. Furthermore, SGCL-DTI provides a new research perspective of contrastive learning for DTI prediction.


AVAILABILITY
The research shows that this method has certain applicability in the discovery of drugs, the identification of drug-target pairs and so on.


SUPPLEMENTARY INFORMATION
Supplementary data are available at Bioinformatics online.

Supervised graph co-contrastive learning for drug-target interaction prediction

dPromoter-XGBoost: Detecting promoters and strength by combining multiple descriptors and feature selection using XGBoost

MOTIVATION
Identification of Drug-Target Interactions (DTIs) is an essential step in drug discovery and repositioning. DTI prediction based on biological experiments is time-consuming and expensive. In recent years, graph learning-based methods have aroused widespread interest and shown certain advantages on this task, where the DTI prediction is often modeled as a binary classification problem of the nodes composed of drug and protein pairs (DPPs). Nevertheless, in many real applications, labeled data are very limited and expensive to obtain. With only a few thousand labeled data, models could hardly recognize comprehensive patterns of DPP node representations, and are unable to capture enough commonsense knowledge, which is required in DTI prediction. Supervised contrastive learning gives an aligned representation of DPP node representations with the same class label. In embedding space, DPP node representations with the same label are pulled together, and those with different labels are pushed apart.


RESULTS
We propose an end-to-end supervised graph co-contrastive learning model for DTI prediction directly from heterogeneous networks. By contrasting the topology structures and semantic features of the drug-protein-pair network, as well as the new selection strategy of positive and negative samples, SGCL-DTI generates a contrastive loss to guide the model optimization in a supervised manner. Comprehensive experiments on three public datasets demonstrate that our model outperforms the SOTA methods significantly on the task of DTI prediction, especially in the case of cold start. Furthermore, SGCL-DTI provides a new research perspective of contrastive learning for DTI prediction.


AVAILABILITY AND IMPLEMENTATION
The research shows that this method has certain applicability in the discovery of drugs, the identification of drug-target pairs and so on.

Supervised graph co-contrastive learning for drug-target interaction prediction.

Background: DNA methylation is an important epigenetic modification, which plays an important role in regulating gene expression at the transcriptional level. In tumor research, it has been found that the change of DNA methylation leads to the abnormality of gene structure and function, which can provide early warning for tumorigenesis. Our study aims to explore the relationship between the occurrence and development of tumor and the level of DNA methylation. Moreover, this study will provide a set of prognostic biomarkers, which can more accurately predict the survival and health of patients after treatment. Methods: Datasets of bladder cancer patients and control samples were collected from TCGA database, differential analysis was employed to obtain genes with differential DNA methylation levels between tumor samples and normal samples. Then the protein-protein interaction network was constructed, and the potential tumor markers were further obtained by extracting Hub genes from subnet. Cox proportional hazard regression model and survival analysis were used to construct the prognostic model and screen out the prognostic markers of bladder cancer, so as to provide reference for tumor prognosis monitoring and improvement of treatment plan. Results: In this study, we found that DNA methylation was indeed related with the occurrence of bladder cancer. Genes with differential DNA methylation could serve as potential biomarkers for bladder cancer. Through univariate and multivariate Cox proportional hazard regression analysis, we concluded that FASLG and PRKCZ can be used as prognostic biomarkers for bladder cancer. Patients can be classified into high or low risk group by using this two-gene prognostic model. By detecting the methylation status of these genes, we can evaluate the survival of patients. Conclusion: The analysis in our study indicates that the methylation status of tumor-related genes can be used as prognostic biomarkers of bladder cancer.

/pdf/identification-of-prognostic-biomarkers-for-bladder-cancer-3im49xk7.pdf

Guohua Wang

Papers

Supervised graph co-contrastive learning for drug-target interaction prediction

dPromoter-XGBoost: Detecting promoters and strength by combining multiple descriptors and feature selection using XGBoost

Supervised graph co-contrastive learning for drug-target interaction prediction.

Identification of Prognostic Biomarkers for Bladder Cancer Based on DNA Methylation Profile

Identifying gene-environment interactions for prognosis using a robust approach.