MS-k NN: protein function prediction by integrating multiple data sources

Abstract

Background Protein function determination is a key challenge in the post-genomic era. Experimental determination of protein functions is accurate, but time-consuming and resource-intensive. A cost-effective alternative is to use the known information about sequence, structure, and functional properties of genes and proteins to predict functions using statistical methods. In this paper, we describe the Multi-Source k-Nearest Neighbor (MS-k NN) algorithm for function prediction, which finds k-nearest neighbors of a query protein based on different types of similarity measures and predicts its function by weighted averaging of its neighbors' functions. Specifically, we used 3 data sources to calculate the similarity scores: sequence similarity, protein-protein interactions, and gene expressions. Results We report the results in the context of 2011 Critical Assessment of Function Annotation (CAFA). Prior to CAFA submission deadline, we evaluated our algorithm on 1,302 human test proteins that were represented in all 3 data sources. Using only the sequence similarity information, MS-k NN had term-based Area Under the Curve (AUC) accuracy of Gene Ontology (GO) molecular function predictions of 0.728 when 7,412 human training proteins were used, and 0.819 when 35,622 training proteins from multiple eukaryotic and prokaryotic organisms were used. By aggregating predictions from all three sources, the AUC was further improved to 0.848. Similar result was observed on prediction of GO biological processes. Testing on 595 proteins that were annotated after the CAFA submission deadline showed that overall MS-k NN accuracy was higher than that of baseline algorithms Gotcha and BLAST, which were based solely on sequence similarity information. Since only 10 of the 595 proteins were represented by all 3 data sources, and 66 by two data sources, the difference between 3-source and one-source MS-k NN was rather small. Conclusions Based on our results, we have several useful insights: (1) the k-nearest neighbor algorithm is an efficient and effective model for protein function prediction; (2) it is beneficial to transfer functions across a wide range of organisms; (3) it is helpful to integrate multiple sources of protein information.

The Ortholog Conjecture Revisited: the Value of Orthologs and Paralogs in Function Prediction

“…The prediction of gene function generally proceeds by the transfer of function from genes with experimental evidence to unannotated, or less-annotated, genes that are similar by some measure [42]. While several methods use multiple data types to carry out predictions [31,47,11], many solely rely on evolutionary relationships [16,24,6,10] and are the focus of the current study.…”

NetGO: Improving Large-scale Protein Function Prediction with Massive Network Information

“…There already exist several attempts on using network information for AFP under the CAFA setting with limited success. For example, MS-kNN, a top method in both CAFA1 and CAFA2, used a weighted average method to integrate multiple sources (information of each source through k-nearest neighbor (KNN)) including networks [24]. However, the accuracy by MS-kNN was similar to that by KNN from a single source, i.e.…”

GOLabeler: Improving Sequence-based Large-scale Protein Function Prediction by Learning to Rank

“…In fact in AFP, several approaches of using the idea of integrating data/classifiers have already been proposed. MS-kNN, a top method in CAFA1 and CAFA2, predicts the function by averaging over the prediction scores from three data sources: sequences, expression and protein-protein interaction [16] (Note that MS-kNN is NOT a sequencebased method). Also Jones-UCL, the top team of CAFA1, integrates prediction scores from multiple methods by using the 'consensus' function (given in Eq.…”