抗微生物肽 (Antimicrobial Peptide)，又稱為宿主防禦胜肽 (Host Defense Peptide)，廣泛存在於各種生物體內，在免疫反應中扮演著重要的角色。抗微生物肽是強效的廣譜 (broad spectrum) 抗生素，生物實驗已證明抗微生物肽可殺死革蘭氏陰性菌、革蘭氏陽性菌、病毒、真菌，甚至是癌細胞，抗微生物肽可以對抗已對現有抗生素產生抗藥性的病原體，因此有望發展為新型治療劑。
近年來抗微生物肽研究有大幅的進展，其中有許多利用機器學習預測抗微生物肽的論文。這些論文大部分是依據胜肽的物理化學特性 (physicochemical properties) 來設計輸入機器學習模型的特徵值，然而這樣的方式需要大量的生醫專業知識，且容易侷限於目前人類對於抗微生物肽機制的理解。隨著深度學習的發展，開始有不少研究將深度學習應用在蛋白質序列上，讓神經網路自行學習蛋白質序列的規律。
本研究提出一個異質卷積神經網路模型以預測抗微生物肽，此模型結合了三個進階卷積神經網路架構，分別是 Deep Residue Network (ResNet)、Densely Connected Network (DenseNet) 以及Squeeze-and-Excitation Network (SENet)，在與其他相關研究比較後，此模型在基準資料集 (Xiao 資料集) 上達到了最好的準確度 (97.6%)、曲線下面積 (99.6%) 以及馬修斯相關係數 (95.2%)。

Antimicrobial peptides (AMPs), also called host defense peptides (HDPs) are part of the innate immune response found among all classes of life. These peptides have a broad spectrum of targets including bacteria, viruses, and fungi. AMPs can kill pathogens that have developed resistance to existing antibiotics. Therefore, antimicrobial peptides demonstrate the potential as novel therapeutic agents.
In recent years, researches concerning AMP prediction have come a long way. Most of the researches take physicochemical properties as input. However, this way highly depends on domain knowledge and may be limited to human understanding of antimicrobial peptides. With the development of deep learning in recent years, many studies applied deep learning to researches on protein sequences. Deep learning models are able to extract important features from raw protein sequences automatically and learn the law of protein sequences.
In this work, we combined three advanced Convolutional Neural Networks (CNNs), including Deep Residue Network (ResNet), Densely Connected Convolutional Network (DenseNet), and Squeeze-and-Excitation Network (SENet), and proposed a heterogeneous CNNs for AMP prediction. When compared to other related works on benchmark dataset (Xiao dataset), the proposed model reached the best accuracy (97.6%), area under ROC curve (99.6%), and Matthews correlation coefficient (95.2%).

1. Chou, K.C., Prediction of protein cellular attributes using pseudo‐amino acid composition. Proteins: Structure, Function, and Bioinformatics, 2001. 43(3): p. 246-255.
2. Lata, S., N.K. Mishra, and G.P. Raghava, AntiBP2: improved version of antibacterial peptide prediction. BMC bioinformatics, 2010. 11(1): p. S19.
3. Thomas, S., et al., CAMP: a useful resource for research on antimicrobial peptides. Nucleic acids research, 2009. 38(suppl_1): p. D774-D780.
4. Xiao, X., et al., iAMP-2L: a two-level multi-label classifier for identifying antimicrobial peptides and their functional types. Analytical biochemistry, 2013. 436(2): p. 168-177.
5. Meher, P.K., et al., Predicting antimicrobial peptides with improved accuracy by incorporating the compositional, physico-chemical and structural features into Chou’s general PseAAC. Scientific reports, 2017. 7: p. 42362.
6. Bhadra, P., et al., AmPEP: Sequence-based prediction of antimicrobial peptides using distribution patterns of amino acid properties and random forest. Scientific reports, 2018. 8(1): p. 1697.
7. Veltri, D., U. Kamath, and A. Shehu, Deep learning improves antimicrobial peptide recognition. Bioinformatics, 2018. 34(16): p. 2740-2747.
8. He, K., et al. Deep residual learning for image recognition. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
9. Huang, G., et al. Densely connected convolutional networks. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.
10. Hu, J., L. Shen, and G. Sun. Squeeze-and-excitation networks. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
11. Dubos, R.J., Studies on a bactericidal agent extracted from a soil bacillus: I. Preparation of the agent. Its activity in vitro. The Journal of experimental medicine, 1939. 70(1): p. 1.
12. McCulloch, W.S. and W. Pitts, A logical calculus of the ideas immanent in nervous activity. The bulletin of mathematical biophysics, 1943. 5(4): p. 115-133.
13. Farley, B. and W. Clark, Simulation of self-organizing systems by digital computer. Transactions of the IRE Professional Group on Information Theory, 1954. 4(4): p. 76-84.
14. Werbos, P., Beyond Regression:" New Tools for Prediction and Analysis in the Behavioral Sciences. Ph. D. dissertation, Harvard University, 1974.
15. LeCun, Y., et al., Gradient-based learning applied to document recognition. Proceedings of the IEEE, 1998. 86(11): p. 2278-2324.
16. Zhou, B., et al. Learning deep features for discriminative localization. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
17. de Paula, V. and A. Valente, A Dynamic Overview of Antimicrobial Peptides and Their Complexes. Molecules, 2018. 23(8): p. 2040.