Transductive multi-label ensemble classification for protein function prediction

IEEE/ACM Trans. Comput. Biol. and Bioinf.

et al. 2013

Self Cite

Abstract-High-throughput experimental techniques produce several kinds of heterogeneous proteomic and genomic datasets. To computationally annotate proteins, it is necessary and promising to integrate these heterogeneous data sources. Some methods transform these data sources into different kernels or feature representations. Next, these kernels are linearly (or non-linearly) combined into a composite kernel. The composite kernel is utilized to develop a predictive model to infer the function of proteins. A protein can have multiple roles and functions (or labels). Therefore, multi-label learning methods are also adapted for protein function prediction.We develop a transductive multi-label classifier (TMC) to predict multiple functions of proteins using several unlabeled proteins. We also propose a method called transductive multi-label ensemble classifier (TMEC) for integrating the different data sources using an ensemble approach. TMEC trains a graph-based multi-label classifier on each single data source and then combines the predictions of the individual classifiers. We use a directed bi-relational graph to capture the relationships between pairs of proteins, between pairs of functions, and between proteins and functions. We evaluate the effectiveness of TMC and TMEC to predict the functions of proteins on three benchmarks. We show that our approaches perform better than recently proposed protein function prediction methods on composite and multiple kernels. The code, datasets used in this paper and supplementary file are available at https://sites.google.com/site/guoxian85/tmec.

show abstract

“…This work is an extension of our earlier paper by Yu et al [21]. In particular, the additional contributions of this paper are as follows.…”

mentioning

confidence: 73%

Section: The Problem Of Undirected Bi-relational Graph In Label Propamentioning

confidence: 99%

Protein Function Prediction Using Multilabel Ensemble Classification

IEEE/ACM Trans. Comput. Biol. and Bioinf.

et al. 2013

Self Cite

show abstract

“…In the bi-relation graph, both proteins and functions are viewed as nodes, and three kinds of edges are defined, namely edges between proteins (exploiting the protein similarity), edges between functions (using function correlations) and edges between function nodes and protein nodes (function annotations). To avoid the risk of functions being overwritten (or missing) in the birelation graph, Yu et al [12] proposed the directed bi-relation graph and applied a random walk with restart [31] on this graph to predict protein functions. Zhang et al [16] used Jaccard coefficients to measure function correlations between different functions and then predicted protein function under a graph-based semi-supervised learning framework [32].…”

Section: Graph-based Protein Function Predictionmentioning

confidence: 99%

“…Some methods use PPI and graph-(or network-) based classifiers to predict the functions of proteins [1], [6], [7], [8], [9]. Several approaches predict protein functions by using heterogeneous data sources (including amino acid sequences and PPI) [2], [10], [11], [12].…”

Section: Introductionmentioning

confidence: 99%

“…Multi-label learning approaches [14], [15] use function correlation to boost the accuracy of protein function prediction [4], [14], [16] and can assign more than one function to a protein. Some protein function prediction methods incorporate the correlations between the functions (labels) to improve the multi-label prediction accuracy [7], [12], [16], [17], [18]. In particular, some approaches first train binary classifiers for each functional label and then utilize the hierarchical structure [19], [20], [21] prevalent within the underlying protein function databases (e.g., Function Catalogue [22] or Gene Ontology [23]).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Protein Function Prediction with Incomplete Annotations

IEEE/ACM Trans. Comput. Biol. and Bioinf.

et al. 2014

Self Cite

Abstract-Automated protein function prediction is one of the grand challenges in computational biology. Multi-label learning is widely used to predict functions of proteins. Most of multi-label learning methods make prediction for unlabeled proteins under the assumption that the labeled proteins are completely annotated, i.e., without any missing functions. However, in practice, we may have a subset of the ground-truth functions for a protein, and whether the protein has other functions is unknown. To predict protein functions with incomplete annotations, we propose a Protein Function Prediction method with W eak-label Learning (ProWL) and its variant ProWL-IF. Both ProWL and ProWL-IF can replenish the missing functions of proteins. In addition, ProWL-IF makes use of the knowledge that a protein cannot have certain functions, which can further boost the performance of protein function prediction. Our experimental results on protein-protein interaction networks and gene expression benchmarks validate the effectiveness of both ProWL and ProWL-IF.

show abstract

Annotating Proteins with Incomplete Label Information

Pattern Recognition in Computational Molecular Biology

2015