Prediction of Selenoproteins Based on Motif Recognition

Abstract

At present available computational methods can not predict selenoproteins correctly because of the special features of selenoproteins. It is known that there are some conservative sections around U in selenoproteins from previous research. So we bring forward a new method to predict selenoproteins based on motif recognition, we use Multiple Em for Motif Elicitation (MEME) to discover motif around U in selenoproteins and then predict selenoproteins based on the motif. The new method found all the selenoproteins in 9 seleno families expect one false positive in family of GPX1 and one in SelS. From the experiment, it is showed that this method can effectively predict almost all the selenoproteins in the known seleno families, and better than the methods of locating the position of U and blasting the Sec/Cys pairing based on handcraft.