Frequent Episode Mining to Support Pattern Analysis in Developmental Biology

Abstract

We introduce a new method for the analysis of heterochrony in developmental biology. Our method is based on methods used in data mining and intelligent data analysis and applied in, e.g., shopping basket analysis, alarm network analysis and click stream analysis. We have transferred, so called, frequent episode mining to operate in the analysis of developmental timing of different (model) species. This is accomplished by extracting small temporal patterns, i.e. episodes, and subsequently comparing the species based on extracted patterns. The method allows relating the development of different species based on different types of data. In examples we show that the method can reconstruct a phylogenetic tree based on gene-expression data as well as using strict morphological characters. The method can deal with incomplete and/or missing data. Moreover, the method is flexible and not restricted to one particular type of data: i.e., our method allows comparison of species and genes as well as morphological characters based on developmental patterns by simply transposing the dataset accordingly. We illustrate a range of applications.

A survey of episode mining

“…The consideration of time constraints has been used in WinMiner (Méger & Rigotti, 2004), EPS, and PPT (Ma et al, 2004) among others. Researchers have also studied the integration of the minimum and maximum number of events per episode (length constraint) for the analysis of heterochrony in developmental biology (Bathoorn et al, 2010).…”