Modeling the next generation sequencing sample processing pipeline for the purposes of classification

Area of Science:

• Genomics
• Systems Biology
• Bioinformatics

Background:

• High-throughput measurements are crucial for developing expression-based classifiers to distinguish cellular phenotypes.
• Next Generation Sequencing (NGS) technologies offer simultaneous gene expression measurements but introduce nonlinear transformations.
• These nonlinearities can reduce data discriminative power compared to direct expression levels.

Purpose of the Study:

• To investigate the impact of the NGS processing pipeline's nonlinear transformation on classification and feature selection.
• To compare NGS-based classification with SAGE-based and raw expression data classification.

Main Methods:

• Utilized state-of-the-art distributional modeling for the NGS processing pipeline.
• Analyzed the effects of various factors on classification accuracy.
• Compared classification performance across NGS, SAGE, and raw expression data.

Main Results:

• The nonlinear transformation inherent in NGS processing diminishes classification accuracy.
• Despite this, NGS-based classification outperforms SAGE-based classification due to a larger number of reads.
• Higher read counts can mitigate performance degradation caused by NGS technologies.

Conclusions:

• High read counts are crucial for maintaining classification performance in NGS analyses.
• Recommends utilizing the highest possible genome coverage during RNA-Seq analysis for optimal classification outcomes.