Classification of methicillin-resistant and methicillin-susceptible staphylococcus aureus using an improved genetic algorithm for feature selection based on mass spectra

Methicillin-resistant staphylococcus aureus (MRSA) cause nosocomial and communal infections seriously. Rapid and accurate identification of MRSA is vital for prevention of human morbidity. Matrix-assisted laser desorption ionization time-offlight mass spectrometry (MALDI-TOF-MS) has been widely used for identification and typing of micro-organisms. To identify MRSA based on the mass spectra of clinical S.aureus, we propose a genetic algorithm with a t-test based population seeding for wrapper feature selection, in which the t-test statistics are used as the prior information for initial population. The results of some compared experiments show that the proposed method improves the average sensitivity from 0.55 to 0.71, and the balanced accuracy is a larger value on contrast group, whose average value is 0.72. As the result, the proposed GA with prior information can identify MRSA effectively.