Phylo-k-mers : a novel approach for sequence analysis. Applications in health and environmental sciences

Phylo-k-mers are phylogenetically informed substrings of length k. From a set of known reference sequence and a tree that represent their relationships (input data), phylo-k-mers can be computed to expand the search space beyond observed sequences on the basis of models of evolution. This 1st phase can be assimilated to a learning phase, in which a set of k-mer associated to tree branches and a probability (features) is stored in a database. In a second step, classifiers can be developed to exploit phylo-k-mers in many applications, from taxonomic or functional classification of biological sequences, to the detection of evolutionary phenomena. After introducing this novel approach, I will discussed its application in phylogenetic placement of environemental data, virus genome recombination detection and classification of proteins into orthologous gene families.