Research On The Prediction Of Minimum Inhibitory Concentration Of Antimicrobial Peptides Based On Sequence Information

With the widespread use of antibiotic drugs,bacteria have gradually resisted to multiple traditional antibiotics.This makes the task of developing new antibacterial drugs imminent.Antimicrobial peptides exist in many species and have a broad spectrum of antibacterial activity.They are small molecular peptides and are used as a part of the immune system to resist infection.Because the resistance mechanism of antimicrobial peptides is different from traditional antibiotics,bacteria hardly resist antimicrobial peptides.Scientists have been devoted to the basic research of antibacterial peptides and development of pertinent drugs.However,researching the properties of antibacterial peptides by biological experiments is time-consuming and costs a lot.Increasingly,researchers study the structural or functional properties of antibacterial peptides depending on bioinformatics.Minimum Inhibitory Concentration(MIC)is considered to be the “gold standard” for measuring the biological activity of antimicrobial peptides.At present,a part of MICs has been measured by biological test.Facing a large number of unidentified MICs of antimicrobial peptides,researchers predict MICs by intelligent computing methods,which will be helpful to the study of antimicrobial peptide.The primary structures of proteins determine the three-dimensional structures and functions,and bioinformatics have developed a series of predictors to forecast various properties of proteins based on sequence information.Xiao et al.constructed a predictor of protein folding rates by incorporating sequence information into sliding window.Inspired by this,after extracting the features based on the antimicrobial peptide sequence information,this paper establishes the predictors of MICs of the antimicrobial peptides which inhibits E.coli and the antimicrobial peptides which inhibit S.aureus by random forests,supports vector machine and the Gaussian kernel regression models.Simulations prove that our predictors forecast MICs easily.Moreover,predictors provide reference values for biologists to measure the MICs by biological test.They help design and reduce the cost of synthesizing novel antimicrobial peptides.Meanwhile,predictors help biologists to develop new gene drugs.