Application of Affinity Propagation on a large breast cancer data set 1

The Affinity Propagation (AP) algorithm is applied to a breast cancer case series subtyping using traditional biological markers. This data set was used to compare the results of AP with the results obtained with standard algorithms. The AP algorithm also provides a procedure to determine the number of profiles to be considered. Results from Affinity Propagation are consistent with the results already obtained having the advantage of providing an indication about the number of clusters. Moreover, results also provide novel insights, with respect to conventional approaches. As a matter of fact, for the Abd El-Rehim et al. data, AP indicates either two, three, four, five, six or eight as an appropriate number of clusters.