FUNCTIONAL SITES IN PRO- AND EUKARYOTIC GENOMES - COMPUTER-MODELS FORPREDICTING ACTIVITY

Here we propose an approach for predicting the activity of functional DNA and RNA sites. This approach includes (1) identification of contex t-dependent conformational, physicochemical, and statistical propertie s of sites significant for their functioning; (2) development of a mod el on their basis for predicting site activity from its sequence; and (3) automatic generation of programs for predicting site activity base d on these models. This approach has been realized as a computer syste m ACTIVITY, which includes databases of site activity as well as confo rmational, physicochemical, and statistical properties of DNA and RNA. ACTIVITY is accessible via Internet (http://www.bionet.nsc.ru/SRCG/Ac tivity/) and allows real-time analysis of experimental data on functio nal site activity. We analyzed 70 samples of sites involved in various molecular biological processes and revealed statistical, conformation al, and physicochemical properties significant for activity of these s ites. We also developed methods for predicting site activity from thei r nucleotide sequences.