Skip to main content
Fig. 7 | BMC Bioinformatics

Fig. 7

From: Protein secondary structure prediction using a small training set (compact model) combined with a Complex-valued neural network approach

Fig. 7

Detailed views of Coil prediction in inhibitors. a Porcine trypsin inhibitor (PDB entry: 1MCT). b cAMP dependent protein kinase inhibitor (PDB entry: 1ATP) with partially visible ATP in yellow. Correct predictions are in light purple and wrong predictions are in magenta. First and last four terminal residues are light brown and are not predicted. N marks the N-terminal. 1ATPI has more correct predictions than 1MCTI. Residues RIWM (5–8) and KCI (19–21) of 1MCTI are Coils wrongly predicted as Sheets. Residues ASGRTGRRN (8–16) of 1ATPI are correct Coil predictions. Waters are red and white sticks in a and red spheres in b. Putative hydrogen bonds (h-bonds) are indicated with dashed black lines, identified by inhibitor polar atom centres within 3.6Å of any O, N atoms. Italics denote the respective enzyme residues (green). The trypsin inhibitor residues make several h-bonds with peptide backbone O, N atoms and the kinase inhibitor, none. Examples in a ARG5 CO with GLY193 NH; ILE6 NH with PHE41 CO. The kinase inhibitor prefers side-chain and water molecule contacts. Examples in b SER9 N with ASP241 OD1; THR12 CO with ARG133 NH1; ARG14 CO with two waters. Not all h-bonds are shown; see text for more

Back to article page