RNA-Seq Count Data Modelling by Grey Relational Analysis and Nonparametric Gaussian Process.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• RNA-sequencing (RNA-seq) generates read count data crucial for gene expression analysis.
• Accurate classification of RNA-seq data is essential for understanding disease pathogenesis and treatment monitoring.
• Existing methods may not fully capture the complexities of RNA-seq read count data.

Purpose of the Study:

• To introduce a novel approach for RNA-seq read count classification.
• To enhance gene feature selection and classification accuracy.
• To provide an effective tool for molecular-level data analysis.

Main Methods:

• Transformation of RNA-seq read counts to microarray-like data for normal-based statistics.
• Grey Relational Analysis (GRA) as an aggregate filter for selecting differentially expressed genes by integrating five individual feature selection methods.
• Nonparametric Bayesian Gaussian Process (GP) model for classification using selected feature subsets.
• Validation using two benchmark real datasets and five-fold cross-validation.

Main Results:

• The GRA-based feature selection method demonstrates superior performance.
• The Gaussian Process (GP) classifier shows dominance over competing methods.
• GRA-GP significantly outperforms sparse Poisson linear discriminant analysis classifiers for read count data.
• The approach is effective across varying numbers of features.

Conclusions:

• The proposed GRA-GP approach offers an effective and robust method for RNA-seq read count data analysis.
• This method enhances the accuracy of gene expression classification.
• It holds significant potential for applications in disease research, diagnosis, and treatment monitoring at the molecular level.