Computational prediction of bacterial type IV-B effectors using C-terminal signals and machine learning algorithms

摘要

Many species of bacteria inject effector proteins to host cells by their type IV secretion systems(T4SS). Two main kinds of T4SS subtypes, IVA and IVB, are well studied in recent years. IVB effectors have been confirmed to be involved in the pathogenicity of various human pathogens. Discriminating these proteins in bacterial genomes are very helpful for identifying their functional roles in hosts. However, there are few effective computational methods can achieve these goals. In this study, the C-terminal sequence features were analyzed, furthermore, a novel algorithm based on machine learning was developed to predict IVB effectors in genomic proteins. Tests on datasets showed that this method can discriminate IVB effectors from non-effectors with over 94.4% accuracy and 81.6% true positive rate. Genome-wide tests in Coxiella burnetii also showed this algorithm is highly sensitive to recognize effector proteins. As a whole, this method is very helpful for new IVB effector identification and other relevant biological studies.