1.Antimicrobial peptides (AMPs) are innate immune components existing in many different organisms. They are promising drug candidates that can overcome drug resistance problems that pose threat to public health. However, identifying AMPs is difficult. To accelerate the process of discovering AMPs, we developed the following technology.
2.We collected more than 6000 AMPs sequence data and then developed a novel protein encoding method to transform data into matrices. We then utilized deep learning to build a classification model to predict whether a protein sequence has antibacterial activity. We set up an online platform as a tool to discern AMPs and non-AMPs. We also developed an AMPs generator using generative adversarial network (GAN) to accelerate the development of new antibacterial drugs.
3.Our deep learning prediction model can reach 90.35% precision. And GAN model can generate sequences highly similar to real AMPs.  
4.Our website: https://symbiosis.iis.sinica.edu.tw/PC_6/

1.We developed a new protein encoding method, PhysicoChemical Property (PC6), which considers both the order and the physicochemical properties of amino acids. This method improves the performance of machine learning. 
2.Our AMPs deep learning prediction model made a breakthrough (precision: 90.35%), which is higher than existing support vector machine (SVM) AMPs prediction models (precision: 85.92%) and random forest (RF) AMPs prediction models (precision: 88.69%).
3.We are the first research team to develop an AMPs generator based on generative adversarial network (GAN). The peptides produced by this generator can highly imitate the amino acid composition ratio and physicochemical properties of existing antibacterial peptides.

1.Market potential and economic benefits: By combining the AMPs prediction platform and the antibacterial peptide generator, pharmaceutical factories and biotechnology companies will be able to obtain candidate protein sequences quickly and predict their antibacterial activities. This can greatly reduce the time and expense for antimicrobial drug development.
2.Applications: Academic Institute, Pharmaceutical Industry, Biotech Company
3.Future development: We plan to synthesize antimicrobial peptides generated and evaluated by our system. Then, we will verify the effect of our generated peptides by biological experiments, hoping to provide a rapid and accurate artificial intelligence-based method for drug developers.