AMP-RNNpro: A two-stage approach for identification of antimicrobials using probabilistic features

Abstract

Abstract Background The necessity to detect antimicrobial peptides (AMPs) using machine learning and deep learning arises from the need for efficiency, accuracy, and the ability to process and analyze large and complex datasets. These tools can complement experimental approaches, accelerate the discovery of AMPs, and contribute to developing effective antimicrobial therapies, especially in the face of increasing antibiotic resistance. Results This study introduced AMP-RNNpro based on Recurrent Neural Network (RNN), an innovative and efficient model for detecting AMPs, which has been constructed based on eight feature encoding methods that are selected according to four criteria: amino acid compositional, grouped amino acid compositional, autocorrelation, and pseudo-amino acid compositional to represent the protein sequences for efficient identification of AMPs by computational methods. In our framework, two-stage predictions have been conducted. At first, an analysis is performed using 33 baseline machine-learning models based on these features. Six models have been selected for further study through performance comparisons using rigorous performance metrics. In the second stage, probabilistic features are generated by deploying these models based on each feature and they are aggregated to be fed into our final meta-model, leading to the precise and time-effective prediction of AMPs. The top 20 features that played greater importance in our model's outcome included AAC, ASDC, and CKSAAGP features that were significantly related to detection and drug discovery. Compared to other state-of-the-art methods, the proposed framework, AMP-RNNpro excels in the indentation of novel AMPs in accuracy and precision, achieving 97.15% accuracy, 96.48% sensitivity, and 97.87% specificity. Conclusions Consequently, our approach can identify AMPs more accurately and rapidly, along with identifying features that could influence the effectiveness of potential treatment discoveries. We built a user-friendly website for the accurate prediction of AMPs based on the proposed approach which can be accessed at AMP-RNNproWebsite.