Ensemble Feature Selection using Rank Aggregation Methods for Population Genomic Data.

Kavakiotis I., Triantafyllidis A., Tsoumakas G., Vlahavas I., (2016) "Ensemble Feature Selection using Rank Aggregation Methods for Population Genomic Data." ACM Proceedings of the 9th Hellenic Conference on Artificial Intelligence, 22, 2016

Bioinformatics, Data Mining, Feature Selection, Rank aggregation, Single Nucleotide Polymorphism, Informative Marker Selection, Population genomics

Single Nucleotide Polymorphisms (SNPs) constitute important genetic markers with numerous medical and biological applications of high scientific and economic interest. SNP datasets are typically high dimensional, containing up to million features. Reasons originating from both biology and machine learning, dictate to perform feature selection which is mainly performed after feature evaluation. In this paper we present methods for SNP evaluation and eventually selection, based on combining results obtained from established genetic marker evaluation methods originating from the field of population genetics. To achieve this we have formulated the feature selection task as a ranking aggregation problem, which is a classical problem in social choice and voting theory.