Objective: The objective of the present work was to investigate key genes in ovarian cancer based on mAP-KL method which comprised the maxT multiple hypothesis (m), Krzanowski and Lai (KL) cluster quality index, and affinity propagation (AP) clustering algorithm, and mutual information network (MIN) constructed by the context likelihood of relatedness (CLR) algorithm.
Materials and methods: MAP-KL method was employed to identify exemplars in ovarian cancer, of which the maxT function ranked the genes of train set and test set and obtained top 200 genes; KL cluster index was utilized to determine the quantity of clusters; and then AP clustering algorithm was conducted to identify the clusters and their exemplars. Also, we assessed the classification performance of mAP-KL by support vector machines (SVM) model. Subsequently, the MIN for exemplars and cluster genes was constructed according to CLR algorithm. Finally, topological centrality properties of exemplars in MIN were assessed to investigate key genes for ovarian cancer.
Results: SVM model validated that the classification between normal controls and ovarian cancer patients by mAP-KL had a good performance. A total of 22 clusters and exemplars were detected by performing the mAP-KL method. Based on the topological centrality analyses for exemplars in MIN, we considered the C9orf16, COX5B and ACTB to be key genes in the progress of ovarian cancer.
Conclusions: We have obtained three key genes (C9orf16, COX5B and ACTB) for ovarian cancer on the basis of mAP-KL method and MIN analysis. These genes might be potential biomarkers for treatment of ovarian cancer, and give insight for revealing the underlying mechanism of this tumor.