Supervised invariant coordinate selection

On linear dimension reduction based on diagonalization of scatter matrices for bioinformatics downstream analyses

Dimension reduction is often a preliminary step in the analysis of data sets with a large number of variables. Most classical, both supervised and unsupervised, dimension reduction methods such as principal component analysis (PCA), independent component analysis (ICA) or sliced inverse regression (SIR) can be formulated using one, two or several different scatter matrix functionals. Scatter matrices can be seen as different measures of multivariate dispersion and might highlight different features of the data and when compared might reveal interesting structures. Such analysis then searches for a projection onto an interesting (signal) part of the data, and it is also important to know the correct dimension of the signal subspace. These approaches usually make either no model assumptions or work in wide classes of semiparametric models. Theoretical results in the literature are however limited to the case where the sample size exceeds the number of variables which is hardly ever true for data sets encountered in bioinformatics. In this paper, we briefly review the relevant literature and explore if the dimension reduction tools can be used to find relevant and interesting subspaces for small-n-large-p data sets. We illustrate the methods with a microarray dataset of prostate cancer patients and healthy controls.

On Exploring Hidden Structures Behind Cervical Cancer Incidence

Finding new etiological components is of great interest in disease epidemiology. We consider time series version of invariant coordinate selection (tICS) as an exploratory tool in the search of hidden structures in the analysis of population-based registry data. Increasing cancer burden inspired us to consider a case study of age-stratified cervical cancer incidence in Finland between the years 1953 and 2014. The latent components, which we uncover using tICS, show that the etiology of cervical cancer is age dependent. This is in line with recent findings related to the epidemiology of cervical cancer. Furthermore, we are able to explain most of the variation of cervical cancer incidence in different age groups by using only two latent tICS components. The second tICS component, in particular, is interesting since it separates the age groups into three distinct clusters. The factor that separates the three clusters is the median age of menopause occurrence.