Prediction of Essential Proteins Using Topological Properties in GO-Pruned PPI Network Based on Machine Learning Methods

The prediction of essential proteins, the minimal set required for a living cell to support cellular life, is an important task to understand the cellular processes of an organism. Fast progress in high-throughput technologies and the production of large amounts of data enable the discovery of essential proteins at the system level by analyzing Protein-Protein Interaction （PPI） networks, and replacing biological or chemical experiments. Furthermore, additional gene-level annotation information, such as Gene Ontology （GO） terms, helps to detect essential proteins with higher accuracy. Various centrality algorithms have been used to determine essential proteins in a PPI network, and, recently motif centrality GO, which is based on network motifs and GO terms, works best in detecting essential proteins in a Baker＇s yeast Saccharomyces cerevisiae PPI network, compared to other centrality algorithms. However, each centrality algorithm contributes to the detection of essential proteins with different properties, which makes the integration of them a logical next step. In this paper, we construct a new feature space, named CENT-ING-GO consisting of various centrality measures and GO terms, and provide a computational approach to predict essential proteins with various machine learning techniques. The experimental results show that CENT-ING-GO feature space improves performance over the INT-GO feature space in previous work by Acencio and Lemke in 2009. We also demonstrate that pruning a PPI with informative GO terms can improve the prediction performance further.

一种蛋白质复合体模块度函数及其识别算法

蛋白质复合体对于研究细胞活动具有重要意义.随着新的生物实验技术的不断出现,产生了大量的蛋白质相互作用网络.通过对蛋白质相互作用网络进行聚类识别蛋白质复合体是当前研究热点.然而,目前大多数蛋白质复合体识别算法的性能不够理想.为此,提出了蛋白质复合体模块度函数(PQ),并在此基础上提出了基于蛋白质复合体模块度函数的模块合并(based on protein complexes modularity function for merging modules,BMM)算法.BMM算法首先识别网络中一些稠密子图作为初始模块,然后依据PQ函数对这些初始模块进行合并,最终得到了质量较高的蛋白质复合体.将识别出的复合体分别与2种已知的蛋白质复合体数据集进行比对,结果表明BMM算法具有很好的识别性能.此外,与其他最新的识别算法相比,BMM算法的识别准确率较高.

A Feature Selection Method for Prediction Essential Protein

Essential proteins are vital to the survival of a cell. There are various features related to the essentiality of proteins, such as biological and topological features. Many computational methods have been developed to identify essential proteins by using these features. However, it is still a big challenge to design an effective method that is able to select suitable features and integrate them to predict essential proteins. In this work, we first collect 26 features, and use SVM-RFE to select some of them to create a feature space for predicting essential proteins, and then remove the features that share the biological meaning with other features in the feature space according to their Pearson Correlation Coefficients（PCC）. The experiments are carried out on S. cerevisiae data. Six features are determined as the best subset of features. To assess the prediction performance of our method, we further compare it with some machine learning methods, such as SVM, Naive Bayes, Bayes Network, and NBTree when inputting the different number of features. The results show that those methods using the 6 features outperform that using other features, which confirms the effectiveness of our feature selection method for essential protein prediction.