Sequence-based analysis and prediction of lantibiotics: A machine learning approach. (December 2018)

• Record Type: Journal Article
• Authors: Poorinmohammad, Naghmeh; Hamedi, Javad; Moghaddam, Mohammad Hossein Abbaspour Motlagh
• Is Part Of: Computational biology and chemistry. Volume 77(2018)
• Page Start: 199
• Page End: 206
• Publication Date: 2018-12
• Subjects: Antimicrobial peptides -- Lanthipeptides -- Support vector machine -- Feature selection -- Peptide design; Chemistry -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; Biochemistry -- Data processing; Biology -- Data processing; Molecular biology -- Data processing; Periodicals; Electronic journals; 542.85
• DOI: 10.1016/j.compbiolchem.2018.10.004
• Languages: English
• ISSNs: 1476-9271

Abstract

Graphical abstract: Highlights: Lantibiotics can be accurately predicted by their sequence via machine learning. SVM and SMO algorithms generated the most accurate lantibiotic predictor models. Feature selection and sequence logo revealed significant distinctions in lantibiotics sequences. The existence of leucine in position 10 from C-termini of most lantibiotics is a meaningful difference. Glutamic acid residue frequency is higher in lantibiotics. Lysin/Arginine residues are mostly distributed in C-termini. Abstract: Lantibiotics, an important group of ribosomally synthesized peptides, represent an important arsenal of novel promising antimicrobials showing high potency in fighting against the prevalence of antibiotic resistance among microbial pathogens. However, due to the lack of high throughput strategies for the isolation and identification of these compounds, our information regarding their structure and especially sequence-based properties is far from complete. Therefore, in the present study, a comprehensive sequence-based analysis of these peptides was performed with the help of machine learning approach together with a feature selection technique. Meanwhile, an attempt to develop an accurate computational model for prediction of lantibiotics was made via constructing two datasets of 280 and 190 lantibiotic and non-lantibiotic antimicrobial peptide sequences, respectively. Based on the conducted approach and as a result of our search for a subset of relevant features of lantibiotics, particular types of sequenced-based features were observed to be preferred in lantibiotics, the knowledge-based implementation of which can be used as strategies for lantibiotic bioengineering purposes. Moreover, a SMO-based classifier was developed for the prediction of lantibiotics with the accuracy and specificity values of 88.5% and 94%, respectively which shows the great potential of the developed algorithm for the prediction of lantibiotcs. Conclusively, the accurate predictor algorithm as well as the identified sequence-based distinctiveness properties of lantibiotics can give valuable information in both the fields of lantibiotic discovery and bioengineering.