Abstract
The accurate identification of anticancer peptides (ACPs) and antimicrobial peptides (AMPs) remains a computational challenge. We propose a tri-fusion neural network termed TriNet for the accurate prediction of both ACPs and AMPs. The framework first defines three kinds of features to capture the peptide information contained in serial fingerprints, sequence evolutions, and physicochemical properties, which are then fed into three parallel modules: a convolutional neural network module enhanced by channel attention, a bidirectional long short-term memory module, and an encoder module for training and final classification. To achieve a better training effect, TriNet is trained via a training approach using iterative interactions between the samples in the training and validation datasets. TriNet is tested on multiple challenging ACP and AMP datasets and exhibits significant improvements over various state-of-the-art methods. The web server and source code of TriNet are respectively available at http://liulab.top/TriNet/server and https://github.com/wanyunzh/TriNet.
[Display omitted]
•Development of TriNet: a deep-learning framework to identify ACPs and AMPs•TriNet captures serial fingerprint and physicochemical distribution features•The framework defines three neural networks to process three peptide features•Development of TVI: an iterative method for training TriNet
In drug discovery, the importance of antimicrobial peptides is increasing as multidrug-resistant microbes continue to emerge. In addition, there is a growing clinical interest in anticancer peptides for the treatment of drug-resistant cancer cells. The cost of traditional wet lab experiments to identify such peptides can be significantly reduced by using computational methods that utilize artificial intelligence. In this study, we developed a deep-learning framework called TriNet for the accurate and rapid identification of anticancer and antimicrobial peptides. Benchmarking studies demonstrate that TriNet performs with extensive adaptability and effectiveness in identifying anticancer and antimicrobial peptides. In this work, TriNet is improved through the appropriate constructions of peptide features, the tri-fusion neural network, and the TVI training method. Further refinement may lead to an effective tool for guiding cancer treatment and antibiotic drug design.
We developed TriNet, a framework for the prediction of anticancer and antimicrobial peptides. The framework captures global sequence and physicochemical distribution information by utilizing three networks, CNN + CAM, Transformer encoder, and Bi-LSTM, which are applied to process these two features and another evolutionary feature. TriNet was trained using TVI, a method developed in this study, and significant improvements are shown in benchmarking studies.