On the use of dissimilarity SOM for DNA barcoding
Résumé
Started in 2007, the international project Barcode of Life is meant to create a large data basis containing most of the existing species and to develop biological and statistical methods, capable of identifying the species of an individual by using a unique genetic "barcode" for all living organisms. This presentation is devoted to the use of clustering methods for the genetic taxonomy problem raised by the project. We will focus on specific Kohonen maps, adapted to this type of data (categorical, high-dimensional, some identical individuals). Several data samples will illustrate the method.