Analysing Cancer Similarities Among Heterogenous Cancer Datatypes Using Similarity Network Fusion

Abstract

Patterns from different datatypes can reveal interesting relationship when they are integrated together. In this research a method of aggregating different data types The similarity network fusion was used over five different cancer data. This technique fuses networks from different cancer data according to their similarities to reveal their patterns. A clustering method called spectral clustering was used to find the patterns in the integrated network. Heatmaps were used to visualise patient to patient relationship and the clusters formed, were the subtypes of patient with similar genetic profile. Similar patient subtypes can be studied for further analys and for medical discoveries.