Dynamic Model for RNA-seq Data Analysis.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• RNA-sequencing (RNA-seq) is a powerful tool for global gene expression profiling.
• RNA-seq data contains noise from measurement errors and biological variation, complicating transcription signal extraction.

Purpose of the Study:

• To develop a robust statistical model for analyzing RNA-seq data.
• To accurately model the underlying transcription process from noisy RNA-seq measurements.
• To apply the model for cancer cell classification and identify cancer-related genes.

Main Methods:

• Utilized a second-order ordinary differential equation (ODE) for modeling RNA-seq data.
• Employed differential principal analysis for estimating location-varying ODE coefficients.
• Validated model accuracy using prediction analysis and 5-fold cross-validation.
• Used ODE model coefficients as features for classifying normal and tumor cells.

Main Results:

• The ODE model accurately fits RNA-seq data.
• High classification accuracy was achieved for distinguishing normal and tumor cells using ODE model features, even for single genes.
• Identified dozens of genes involved in cancer through response analysis to external signal perturbations.

Conclusions:

• The proposed ODE model effectively extracts biologically relevant transcription signals from RNA-seq data.
• The model demonstrates significant potential for cancer diagnostics and understanding transcription dynamics in disease.