Classifying leukemia types with chromatin conformation data
- Equal contributors
1 Department of Biochemistry and Goodman Cancer Research Center, McGill University, Montréal, Québec H3G 1Y6, Canada
2 McGill Centre for Bioinformatics and School of Computer Science, McGill University, Montréal, Québec H3G 0B1, Canada
Genome Biology 2014, 15:R60 doi:10.1186/gb-2014-15-4-r60Published: 30 April 2014
Although genetic or epigenetic alterations have been shown to affect the three-dimensional organization of genomes, the utility of chromatin conformation in the classification of human disease has never been addressed.
Here, we explore whether chromatin conformation can be used to classify human leukemia. We map the conformation of the HOXA gene cluster in a panel of cell lines with 5C chromosome conformation capture technology, and use the data to train and test a support vector machine classifier named 3D-SP. We show that 3D-SP is able to accurately distinguish leukemias expressing MLL-fusion proteins from those expressing only wild-type MLL, and that it can also classify leukemia subtypes according to MLL fusion partner, based solely on 5C data.
Our study provides the first proof-of-principle demonstration that chromatin conformation contains the information value necessary for classification of leukemia subtypes.