Quantitatively Assessing the Effects of Regulatory Factors on Nucleosome Dynamics by Multiple Kernel Learning

Nucleosome, a nucleoprotein structure formed by coiling 147bp of DNA around an octamer of histone proteins, is the fundamental repeating unit of eukaryotic chromatin. By regulating the access of biological machineries to underlying textit{cis}-regulatory elements, its mobility has been implicated in many important cellular processes. Although it has been known that various factors, such as DNA sequences, histone modifications, etc., cooperatively affect nucleosome mobility, the contribution of each factor in the common impact remains unclear. We propose, in this work, a novel computational approach based on Multiple Kernel Learning (MKL) for quantitatively assessing the effects of two important factors, i.e., genomic sequence and histone modifications, on nucleosome dynamics. Our result on textit{S.cerevisiae} shows that, epigenetic feature, such as histone modifications, plays more important role than genomic sequence in regulating nucleosome dynamics.