Séquences Classification

The goal of the DNA_Classifier is to categorize data sequences for a patient. This method is based on sequential patterns (see Figure 1). With the DNA_Classifer, it is possible to predict the grade of a cancer for a patient (see Figure 2).


Figure 1 – sequence visualisation

This prototype is based on the following principles. For each patient, a DNA microarray is produced and the analysis of gene expression is used to analyze a particular disease. From the genes expression, it is possible to define sequences of data depending on the value of the expression of the gene. For example, the data sequence <(5) (8 9) (10)> could be interpreted as follows: the gene 5 has a lower expression than the gene 8 which have the same value than the gene 9. Both genes 8 and 9 have an expression lower than gene 10. DNA_Classifier offers three different ways of classifying sequential patterns in order to facilitate the selection of sequential patterns the most characteristic of a class of tumor.

The results obtained with DNA_Classifier resulted in a publication in the Journal of Biomedical Informatics in 2011.

Patterns classification

Figure 2- Classification results

Last update on 20/01/2015