Membrane proteinis an important kind of proteins embedded in the membranes of cells and play crucial roles in living organisms, such as ion channels, transporters, receptors. Because it is difficult to determinate the membrane protein’s structure by wet-lab experiments, accurate and fast amino acid sequence-based computational methods are highly desired. In this paper, we developed an online prediction tool called MemBrain, whose input is the amino acid sequence. MemBrain consists of specialized modules for predicting transmembrane helices, residue-residue contacts and relative accessible surface area of α-helical membrane proteins. MemBrain achieves a prediction accuracy of 97.9% of A_TMH, 87.1% of A_P, 3.2±3.0 of N-score, 3.1±2.8 of C-score. MemBrain-Contact obtains 62%/64.1% prediction accuracy on training and independent dataset on top 5/L contact prediction respectively. And MemBrain-Rasa achieves pearson correlation coefficient of 0.733 and its mean absolute error of 13.593. These prediction results are higher than that of the existing methods and provide valuable hints for revealing the structure and function of membrane proteins. MemBrain web-server is free academic use available at www.csbio.sjtu.edu.cn/bioinf/MemBrain/.

膜蛋白是嵌入细胞膜中的一种重要蛋白质，在离子通道，转运蛋白，受体等生物体中起着至关重要的作用。由于很难通过湿实验室实验确定膜蛋白的结构，因此迫切需要基于氨基酸序列的准确，快速的计算方法。在本文中，我们开发了一种称为MemBrain的在线预测工具，其输入是氨基酸序列。MemBrain由专门的模块组成，用于预测跨膜螺旋，残基-残基接触和α-螺旋膜蛋白的相对可及表面积。MemBrain的预测准确度达到A _TMH的97.9 ％，A _P的87.1％，N得分3.2±3.0，C得分3.1±2.8。MemBrain-Contact在训练上获得62％/ 64.1％的预测准确度，在前5 / L个接触预测上分别获得独立的数据集。MemBrain-Rasa的皮尔逊相关系数为0.733，平均绝对误差为13.593。这些预测结果高于现有方法的预测结果，并为揭示膜蛋白的结构和功能提供了有价值的提示。MemBrain网络服务器可从www.csbio.sjtu.edu.cn/bioinf/MemBrain/免费获得学术使用。