Rapid identification of key genes for the rod-shaped morphology in bacteria using multi-species genomes

Abstract

AbstractAccurate identification of key genes is pivotal in biological research. Here, we introduce machine learning to the field of functional gene identification, enabling precise prediction of bacterial shape based on genomic information. Our machine learning model successfully predicts bacterial shape, and we determine the influence of various protein domains on shape using the model. This approach facilitates the identification of candidate genes involved in regulating bacterial shape. Through targeted knockout experiments on eight potential key regulatory genes (pal, yicC, mreB, tolQ, ftsX, amiC, yddB, andrpoZ) inEscherichia coli, we observe significant alterations in rod-shaped morphology upon individual knockout ofpalandmreBgenes.E. colitransitions from rod-shaped to spherical or cell wall-deficient protoplasmic states. Experimental validations validate the robustness of our newly developed method. This study establishes an innovative avenue for exploring functional genes, harnessing large-scale genomic information to promptly uncover key genes governing shared traits across species.