Through the Ramachandran Haze: Ca-Parameters Reveal Secondary Structure at Low Resolution

Abstract

CaBLAM (Calpha-Based Low-resolution Annotation Method) addresses the challenge of producing stereochemically accurate models from rough chain tracings at low resolution (3-4A). Much current excitement in biological crystallography centers on large complexes and “molecular machines”, but the resulting models can be subject to artifacts arising from inherent properties of low-resolution electron density maps.Even at poor resolution, chain tracing can produce full backbone models, achieving the primary objective of locating the amino-acid residues in 3D space. However, structural details like peptide orientation are very often distorted by misleading or ambiguous density. Mismodeled backbone wreaks havoc with standard ways of identifying secondary structure, placing side chains, and refining a reasonable model. CaBLAM uses carbonyl oxygen placement to diagnose commonly occurring patterns of correctable backbone errors, then uses contours derived from a high-quality dataset in a novel parameter space of overlapping Cα pseudo-dihedrals to identify the secondary structures disguised by those errors.In addition to continuous secondary structures like alpha helix and beta sheet, CaBLAM can identify non-continuous secondary structures such as helix caps, tight turns, and beta bulges, even in low-resolution models. The power to distinguish between modeling errors and these real irregularities will provide refinement with a more realistic, more detailed protein backbone from which to produce a reliable structural model.