With the development of science and biotechnology, many evidences show that ncRNAs play an important role in the development of important biological processes, especially in chromatin modification, cell differentiation and proliferation, RNA progressing, human diseases, etc. Moreover, lncRNAs account for the majority of ncRNAs, and the functions of lncRNAs are expressed by the related RNA-binding proteins. It is well known that the experimental verification of lncRNA-protein relationships is a waste of time and expensive. So many time-saving and inexpensive computational methods are proposed to uncover potential lncRNA-protein interactions. In this work, we propose a novel computational method to predict the potential lncRNA-protein interactions with the bipartite network projection recommended algorithm (LPI-BNPRA). Our approach is a semi-supervised method based on the lncRNA similarity matrix, protein similarity matrix, and lncRNA-protein interaction matrix. Compared with three previous methods under the leave-one-out cross-validation, our model has a more high-confidence result with the AUC value of 0.8754 and the AUPR value of 0.6283. We also do case studies by the Mus musculus dataset to further reflect the reliability of our approach. This suggests that LPI-BNPRA will be a reliable computational method to uncover lncRNA-protein interactions in biomedical research.

随着科学和生物技术的发展，许多证据表明ncRNA在重要的生物学过程的发展中起着重要作用，尤其是在染色质修饰，细胞分化和增殖，RNA进展，人类疾病等方面。此外，lncRNA占了大多数。 ncRNA的表达，而lncRNA的功能由相关的RNA结合蛋白表达。众所周知，lncRNA-蛋白质关系的实验验证是浪费时间和昂贵的。提出了许多节省时间和廉价的计算方法来发现潜在的lncRNA-蛋白质相互作用。在这项工作中，我们提出了一种新的计算方法，以预测与双向网络投影推荐算法（LPI-BNPRA）的潜在lncRNA-蛋白质相互作用。我们的方法是基于lncRNA相似性矩阵，蛋白质相似性矩阵和lncRNA-蛋白质相互作用矩阵的半监督方法。与留一法交叉验证下的三种先前方法相比，我们的模型具有更高的置信度，AUC值为0.8754，AUPR值为0.6283。我们还通过小家鼠数据集进一步体现了我们方法的可靠性。这表明LPI-BNPRA将是发现生物医学研究中lncRNA-蛋白质相互作用的可靠计算方法。