Regression as a method to predict copy numbers in comparative genomic hybridization studies on bacteria

Area of Science:

• Genomics
• Microbiology
• Bioinformatics

Background:

• Comparative genomic hybridizations (CGH) using microarrays assess genomic similarity between bacterial strains.
• Microarrays are based on a reference strain genome, with unknown strains compared against it.
• Fluorescent intensity ratios reveal gene divergence and predict gene copy number in unknown strains.

Purpose of the Study:

• To develop and evaluate novel methods for predicting bacterial gene copy number from CGH data.
• To compare the performance of proposed predictors against direct intensity ratio methods.

Main Methods:

• Assumed a linear relationship between log2 copy number and observed log2-ratios.
• Developed predictors based on factor analysis and linear random models.
• Applied proposed methods to CGH data from Enterococcus faecalis strains.

Main Results:

• Simulations demonstrated that the proposed predictors generally improved gene copy number prediction accuracy.
• The methods were successfully applied to real CGH data from five Enterococcus faecalis strains.
• Identified copy numbers of relevant genes in the tested strains.

Conclusions:

• The proposed factor analysis and linear random models offer enhanced accuracy for predicting bacterial gene copy number from CGH experiments.
• These improved prediction methods are valuable for comparative genomics and microbial strain analysis.