PepLook Scale-Up Prediction of protein structures

Besides experimental approaches for peptide structures determination which results often differ with assay conditions, there is, to our knowledge, only three in silico methods available for the prediction of peptide structures: Pepstruct, Rosetta and PepLook. The latter was developed by the CBMN (1, 2) based on the fact that any protein PDB model can be re-constructed in silico using a restricted subset of φψ couples of angles (3). As PepLook was able to predict conformation of peptides and protein fragments, like cell penetrating peptides (CPPs) (1) and the hydrophobic segment of DGKє (4) with good accuracy, we are testing whether PepLook could be used for the prediction of complete protein structures. To reach this goal, PepLook is used to predict the conformation of peptidic fragments along the protein sequences. The sequence is scanned with different sizes of windows shifted along the sequence from the first to the last residue. For each sequence window, the 99 PepLook models of structure are analysed and compared to the PDB model. PepLook scans are running for five proteins, α-synuclein (1XQ8), a Zinc endoprotease (1C7K), Ubiquitin (1UBQ), Cytochrome b562 (256B) and Lysozyme (3LZT, 1AM7), using different window lengths (7, 9, 11, 15, 17, 21, 23 and 27 residues) by sliding steps of 1 residue. Since this approach requires huge calculation time, we present here the preliminary results.