Shape modeling and matching in identifying protein structure from low-resolution images

Abstract

Figure 1: Identifying α-helices in a low-resolution protein image, using the Human Insulin Receptor -Tyrosine Kinase Domain (1IRK) as an example. The inputs are the amino-acid sequence of the protein (a), where α-helices are highlighted in green, and a density volume reconstructed from electron cryomicroscopy (b), where possible locations of α-helices have been detected as cylinders shown in (c). Our method computes the correspondence between the helices in the sequence and in the density volume (e). This is achieved by extracting a skeleton from the density volume shown in (d) and matching it with the sequence in (a). Note that the matching is error-tolerant therefore the resulting correspondence does not have to be a bijection. AbstractIn this paper, we describe a novel, shape-modeling approach to recovering 3D protein structures from volumetric images. The input to our method is a sequence of α-helices that make up a protein, and a low-resolution volumetric image of the protein where possible locations of α-helices have been detected. Our task is to identify the correspondence between the two sets of helices, which will shed light on how the protein folds in space. The central theme of our approach is to cast the correspondence problem as that of shape matching between the 3D volume and the 1D sequence. We model both the shapes as attributed relational graphs, and formulate a constrained inexact graph matching problem. To compute the matching, we developed an optimal algorithm based on the A*-search with several choices of heuristic functions. As demonstrated in a suite of real protein data, the shape-modeling approach is capable of correctly identifying helix correspondences in noise-abundant volumes with minimal or no user intervention.

Protein Data Representation: A Survey

“…This error is typically the cost of deforming the original graphs to their subgraphs and the error of matching the attributes of corresponding elements in the two subgraphs. The most popular approach for error-correcting graph matching is the A* algorithm [32].…”

“…Abeysinghe et al [32] introduced a representation using Attributed Relational Graphs (ARG).It presented an application for identifying protein structure from images. It represented the shapes of biological data (e.g., protein sequence and density volume) as ARG.…”

Coarse-to-fine skeleton extraction for high resolution 3D meshes

“…The skeleton has proven to be a valuable and widely used shape descriptor for a number of tasks such as 2-D and 3-D shape recognition [2,3], volumetric models deformation [4,5], segmentation [6] and protein structure identification [7]. The interest in this descriptor stems from its being a concise representation of the original shape, which is topologically equivalent to it, and invariant to several shape deformations.…”

An Adaptive Hierarchical Approach to the Extraction of High Resolution Medial Surfaces

“…The skeleton has proven to be a valuable and widely used shape descriptor for a number of tasks such as 2-D and 3-D shape recognition [13,19], volumetric models deformation [21], segmentation [15] and protein structure identification [1]. When working in two dimensions, the skeleton, or medial axis transform, is defined as the locus of the centers of the maximal inscribed circles bitangent to the shape boundary.…”