I'm working on a project where I'm using similar methods, but not from Ayasdi, to study cyclic phenomena in high-dimensional data.