engleski

Bioinformatics study of nuclear receptor superfamily relation to UniProt data

Nuclear receptors (NR) classification is based on sequence comparison for conserved domains responsible for the DNA binding and the ligand binding, (C and E domain, respectively.) This well developed system for classification inside the superfamily of nuclear receptors can not be used to relate NR to the rest of protein universe, but we developed a general classification method based on size invariant represenation of the sequence which can do that. The size invariant representation used is of adjacency histogram type with user controlled amount of information, which makes the method of adaptable resolution in space between seqence and motif based classifications. The adjacency histogram is constructed using controlled number and positions of neigbours in the sequence. Inside this representation the neigbours can be amino acids themselves or some combination of their characteristic physicochemical properties, like acidity, hidrofobicity, polarity, aromacity, size, etc. In the first step, the classification parameters were optimized to repeat the results of sequence based classification inside the superfamily of nuclear receptors. The analyis and optimization of representations are done using eigenanalyis and clustering methods. The same representations are then constructed for proteins from UniProt data base and relation of nuclear receptors to the rest of the universe was analysed finding closest negbours to nuclear receptor superfamily.

Nuclear receptors; bioinformatics; classification; eigenanalysis; clustering

nije evidentirano