Parallel Affinity Propagation Clustering in Identifying Sub-Network Biomarker Genes of Lung Cancer

Lung cancer is a complex disease. The identification of a disease biomarker is still challenging. Affinity Propagation (AP) is a clustering algorithm to cluster a set of data by identifying similar data points in an iterative process. By applying a microarray dataset, this leads to a scalability issue for large data points. In this work, Pearson's correlation was used for calculating a similarity matrix with subsequent pruning for further constructing a gene co-expression network. The AP has been applied to identify sub-network biomarkers of four lung cancer expression datasets based on two different microarray platforms. Parallel computing was applied to tackle high dimensionality and to reduce the time consumption of measuring similarity by Pearson's correlation and similarity matrix construction.