ABSTRACT
In this chapter an application of biclustering in genetics is demonstrated. Using FABIA from Chapter 8, the HapFABIA method finds biclusters in genotyping data and, thereby, identifies DNA regions that are shared between individuals. DNA regions are shared between different human populations, between humans and Denisovans, and between humans and Neandertals. This chapter is based on previous publications: Hochreiter et al. (2010); Hochreiter (2013); Povysil and Hochreiter (2014). FABIA and HapFABIA use variational approaches to biclustering (Hochreiter et al., 2010, 2006; Klambauer et al., 2012, 2013). The FABIA learning procedure is a variational expectation maximization (EM), which maximizes the posterior of the parameters (Hochreiter et al., 2010; Talloen et al., 2010; Hochreiter et al., 2006; Talloen et al., 2007; Clevert et al., 2011; Klambauer et al., 2012, 2013). In the next section we ex-
plain “identity by descent” (a genetic relationship between individuals), then we demonstrate how it is found by biclustering, then we show how FABIA biclustering is used to detect DNA regions of identity by descent, and finally we present some case studies on genetic data.
