Toward a multilevel representation of protein molecules: Comparative approaches to the aggregation/folding propensity problem

Abstract

This paper builds upon the fundamental work of Niwa et al. [34], which provides the unique possibility to analyze the relative aggregation/folding propensity of the elements of the entire Escherichia coli (E. coli) proteome in a cell-free standardized microenvironment. The hardness of the problem comes from the superposition between the driving forces of intra-and inter-molecule interactions and it is mirrored by the evidences of shift from folding to aggregation phenotypes by single-point mutations [10]. Here we apply several state-of-the-art classification methods coming from the field of structural pattern recognition, with the aim to compare different representations of the same proteins gathered from the Niwa et al. data base; such representations include sequences and labeled (contact) graphs enriched with chemico-physical attributes. By this comparison, we are able to identify also some interesting general properties of proteins. Notably, (i) we suggest a threshold around 250 residues discriminating "easily foldable" from "hardly foldable" molecules consistent with other independent experiments, and (ii) we highlight the relevance of contact graph spectra for folding behavior discrimination and characterization of the E. coli solubility data. The soundness of the experimental results presented in this paper is proved by the statistically relevant relationships discovered among the chemico-physical description of proteins and the developed cost matrix of substitution used in the various discrimination systems.

Multifractal cross-correlation effects in two-variable time series of complex network vertex observables

“…This stresses the novelty of our approach with respect to previous single-variable studies [15][16][17]. We discuss the results of the simulations performed on both network models and real networks representing folded proteins [26]. Our results suggest that the proposed analysis framework is useful to investigate the multi-level organization of networks from the perspective of coupling between two different topological features of vertices.…”

“…In the literature, such networks are typically called protein contact networks (PCNs). Here we process two sample PCNs (in the following denoted as JW0058 and JW0179) taken from [16,26].…”

One-Class Classifiers Based on Entropic Spanning Graphs

“…This data representation is obtained by "seriating" the graphs. Therefore, a sequence contains a number of elements equal to the number of vertices in the related graph; see [62] for details on the seriation algorithm. In this case, d I (·, ·) is implemented by using the dynamic time warping algorithm, equipped with a weighted Euclidean distance for computing the distances between the three-dimensional vectors composing the sequences.…”

A generative model for protein contact networks

“…The first ensemble of graphs contains PCNs, directly obtained from the 3D native structures resolved for the E. coli proteome [36,37]. Each vertex is defined as the alpha carbon of a residue; edges are added among two residues if their Euclidean distance is within the [4,8]Å range.…”

“…The contribution of this paper consists in a two-step generative model for PCNs; the first stage of our method takes inspiration from the work of Bartoli et al [5]. The dataset considered in our study consists of four ensembles (classes) of networks: i) actual PCNs elaborated from the E. coli proteome [36,37], ii) synthetic networks generated according to the recipe of Bartoli et al [5], iii) synthetic modular networks generated with the method proposed by Sah et al [57], and finally iv) those generated with our method. We evaluate the soundness of the proposed approach by focusing on mesoscopic analyses.…”