Convolutional neural network-based annotation of bacterial type IV secretion system effectors with enhanced accuracy and reduced false discovery

• DOI: 10.1093/bib/bbz120
• Publication Date: 2019-08-22T11:38:12Z
• Authors: Jiajun Hong; Yongchao Luo; Minjie Mou; Jianbo Fu; Yang Zhang; Weiwei Xue; Tian Xie; Lin Tao; Yan Lou; Feng Zhu
• Keywords: Benchmark (surveying)

The type IV bacterial secretion system (SS) is reported to be one of the most ubiquitous SSs in nature and can induce serious conditions by secreting SS effectors (T4SEs) into host cells. Recent studies mainly focus on annotating new T4SE from huge amount sequencing data, various computational tools are therefore developed accelerate annotation. However, these as heavily dependent selected methods their annotation performance need further enhanced. Herein, a convolution neural network (CNN) technique was used annotate T4SEs integrating multiple protein encoding strategies. First, accuracies nine strategies integrated with CNN were assessed compared that popular based independent benchmark. Second, false discovery rates models systematically evaluated (1) scanning genome Legionella pneumophila subsp. ATCC 33152 (2) predicting real-world non-T4SEs validated using published experiments. Based above analyses, strategies, (a) position-specific scoring matrix (PSSM), (b) secondary structure & solvent accessibility (PSSSA) (c) one-hot scheme (Onehot), identified well-performing when CNN. Finally, novel strategy collectively considers three (CNN-PSSM, CNN-PSSSA CNN-Onehot) proposed, tool (CNN-T4SE, https://idrblab.org/cnnt4se/) constructed facilitate All all, this study conducted comprehensive analysis collection CNN, which could suppression T4SS infection limit spread antimicrobial resistance.