A machine learning approach to identifying Salmonella stress response genes in isolates from poultry processing

Area of Science:

• Microbiology
• Genomics
• Bioinformatics

Background:

• Salmonella contamination in poultry processing poses a significant food safety risk.
• Understanding Salmonella stress response mechanisms is crucial for developing effective control strategies.
• Whole genome sequencing (WGS) offers a powerful tool for genetic analysis of bacterial pathogens.

Purpose of the Study:

• To apply machine learning algorithms to Salmonella WGS data for identifying significant genes associated with stress response during poultry processing.
• To evaluate the performance of various machine learning models in predicting Salmonella stress-related genes.
• To pinpoint specific genes involved in Salmonella adaptation to poultry processing environments.

Main Methods:

• Utilized whole genome sequencing (WGS) data from 177 Salmonella isolates obtained from various poultry processing stages.
• Trained six machine learning algorithms: random forest, neural network, cost-sensitive learning, logit boost, and support vector machine (linear and radial kernels).
• Assessed model performance using area under the receiver operating characteristic (AUROC) curve, sensitivity, and specificity.

Main Results:

• All machine learning models demonstrated high performance, with logit boost achieving the best AUROC score of 0.904.
• Identified significant genes including ybtX (zinc transporter), yccK and thiS (transferase-encoding), and cold/heat shock genes (cspA, cspD, cspE, rpoH, rpoE).
• Other identified genes are involved in lipopolysaccharide biosynthesis, DNA repair, biofilm formation, and cellular metabolism.

Conclusions:

• Machine learning effectively identifies genes critical to Salmonella stress response in poultry processing.
• The identified genes provide potential targets for mitigating Salmonella contamination in poultry.
• This study demonstrates the utility of WGS combined with machine learning for pathogen surveillance and control.