Combining fold recognition and exploratory data analysis for searching for glycosyltransferases in the genome of...

Fold recognition was applied to the systematic analysis of the all sequences encoded by the genome of Mycoplasma tuberculosis H37Rv in order to identify new putative glycosyltransferases. The search was conducted against a library composed of all known crystal structures of glycosyltransferases and some related proteins. A clear relationship appeared between some sequences and some folds. It appears necessary to complete the fold recognition approach with a statistical approach in order to identify the relevant data above the background noise. Exploratory data analysis was carried out using several methods. Analytical methods confirmed the validity of the approach, while predictive methods, although very preliminary in the present case, allowed for identifying a number of sequences of interest that should be further investigated. This new approach of combining bioinformatics and chemometrics appears to be a powerful tool for analysis of newly sequenced genomes. Its application to glycobiology is of great interest.