Simultaneous identification of robust synergistic subnetwork markers for effective cancer prognosis

Abstract

Background: Accurate prediction of cancer prognosis based on gene expression data is generally difficult, and identifying robust prognostic markers for cancer remains a challenging problem. Recent studies have shown that modular markers, such as pathway markers and subnetwork markers, can provide better snapshots of the underlying biological mechanisms by incorporating additional biological information, thereby leading to more accurate cancer classification. Results: In this paper, we propose a novel method for simultaneously identifying robust synergistic subnetwork markers that can accurately predict cancer prognosis. The proposed method utilizes an efficient message-passing algorithm called affinity propagation, based on which we identify groups -or subnetworks -of discriminative and synergistic genes, whose protein products are closely located in the protein-protein interaction (PPI) network. Unlike other existing subnetwork marker identification methods, our proposed method can simultaneously identify multiple nonoverlapping subnetwork markers that can synergistically predict cancer prognosis. Conclusions: Evaluation results based on multiple breast cancer datasets demonstrate that the proposed message-passing approach can identify robust subnetwork markers in the human PPI network, which have higher discriminative power and better reproducibility compared to those identified by previous methods. The identified subnetwork makers can lead to better cancer classifiers with improved overall performance and consistency across independent cancer datasets.

Incorporating topological information for predicting robust cancer subnetwork markers in human protein-protein interaction network

“…The parameter, α , is defined between [0,1] to control this term. It is shown in our previous work [ 11 ] that the size of the network decreases as α gets larger. It is because a larger α tends to cluster genes with similar discriminative power.…”

“…We adopt the subnetwork identification procedure from our previous study [ 11 ], where we utilized a message-passing clustering algorithm, called affinity propagation, to cluster genes whose protein products interact with each other or are closely located in PPI network. The input of this clustering algorithm is the measure of similarity between genes.…”

“…We use the same setting for preference as in [ 11 ]. The self-similarity is set to s ( k,k )= c for all k , where s ( i,k )≤ c for only 1 % of all gene pairs ( g i , g k ) to guarantee that every gene gets equal chance to be an exemplar at the initial stage of clustering process.…”

“…As cancer is a complex disease which its progression involves dysregulation of multiple genetic processes, there is an alternative approach based on the assumption that genes which are known to be in common pathways [ 4 – 8 ] or genes whose protein products are functionally related in protein-protein interaction (PPI) networks [ 9 – 11 ] should be interpreted together as a single feature. This approach analyzes gene expression data at “modular” level by integrating biological information, such as known molecular pathways or PPI networks.…”

“…Su et al [ 10 ] proposed dynamic programming method to identify and greedily combined paths containing differentially expressed and coexpressed genes to obtain subnetwork markers for predicting breast cancer metastasis. More recently, in our previous work [ 11 ], we utilized a message-passing clustering algorithm to identify subnetwork markers with high-accuracy disease prediction. The method is capable to simultaneously predict multiple non-overlapping subnetwork markers which may lead to cover more genes with lower computational cost compared to the existing methods.…”