Kinemage visualizations of AD assignment of secondary structure, and a comparison table between AD and DSSP assignments are on this page. See also our kinemages showing work in progress on hinge detection using AD tetrahedra in TrpRS.

Visualization of AD tetrahedral patterns in a synthetic alpha-helix (download Kinemage).

• More information about the Kinemage 3D visualization format is here, and the Mage program for various platforms may be downloaded here.

Background and Tutorial

Click here for a general context about almost-Delaunay tetrahedra and their other uses in the analysis of protein structure.

These kinemages depict the AD secondary structure assignment method, which is described on our project page and in detail in an upcoming paper. We compare the AD assignment of secondary structure with the widely used DSSP assignments and with a visual-geometric method that we implemented (similar to GAS-P).

Observations  

Below are some screen shots of the visualization. Click on each image for a high-resolution image and on the "kinemage" link for the corresponding Kinemage animation. In these kinemages, the C-α trace is colored red (for α-helices), green (β-sheets), purple (β-turns), and white (unassigned). We may animate between or overlay the AD, DSSP and visual-geometric assignments to compare them.

2bnh, 456 residues (download kinemage)	1mbn, 153 resiues (download kinemage)
1ospO, 251 residues (download kinemage)	1bg5, 254 residues (download kinemage)

Conclusions   We observe that α-helices assigned by AD match in their number and their termini with both DSSP and visual-geometric methods, but where they do not match with DSSP (eg. the helix in 1bg5, residues 118-133, that DSSP assigns as β-turns), they match at least partially with the visual-geometric assignment. For β-sheets, there are several cases where the current method is inaccurate compared to DSSP, since AD thresholds do not give a signature for β-sheet, and the method uses only the C-α coordinates. But in several cases our method seems to detect β-sheets where DSSP does not because of non-robustness of its H-bond detection or irregular H-bonding patterns. β-turns are shown compared with DSSP, whereas PROMOTIF definitions (based on phi-psi angles) are more accurate; future versions of these kinemages will compare with PROMOTIF, while the comparison table already does.

Project Members  

Deepak Bandyopadhyay Graduate Student UNC Chapel Hill debug at cs.unc.edu

Alexander Tropsha Professor and Director, Laboratory of Molecular Modeling, School of Pharmacy, UNC Chapel Hill alex_tropsha at unc.edu

Charles W. Carter, Jr. Professor of Biochemistry and Biophysics, School of Medicine, UNC Chapel Hill carter at med.unc.edu

Jack Snoeyink Professor of Computer Science UNC Chapel Hill snoeyink at cs.unc.edu