Fig. 1: Overall pipeline for the GA–AutoML workflow and biological validation in P. aeruginosa. | npj Systems Biology and Applications

Fig. 1: Overall pipeline for the GA–AutoML workflow and biological validation in P. aeruginosa.

From: Minimal gene signatures enable high-accuracy prediction of antibiotic resistance in Pseudomonas aeruginosa

Fig. 1

A Incorporation of RNA-seq data from 414 clinical isolates alongside antibiotic phenotypes and initial data processing. B Application of a genetic algorithm that iteratively identifies minimal, high-accuracy gene subsets through population initialization, selection, crossover, and mutation steps. C Integration of these GA-selected features into data preprocessing, classification, model selection, and final performance evaluation under an automated machine-learning pipeline. D Post hoc validation of selected genes through multiple external resources (CARD, Pseudomonas Genome DB, ShinyGO, NCBI, iModulonDB), with a focus on their operon assignments, potential regulatory networks, and broader functional context19,20,60,63,64. Created in BioRender. Saha, R. (2025) https://BioRender.com/0e3kj5x.

Back to article page