Remote homology detection of pr oteins using 3D models enriched with physicochemical pr operties.

(Eng) This article introduces a new method for remote homology detection called remote-3DP. He remote 3DP method is based on both predicted 3D information and the physicochemical properties of the amino acids. The method considers only 10 structural models to represent a protein and distinguish the remote counterparts of non-remote ones in 54 SCOP families. The low dimensionality of the representation allows use different classification techniques and find out which one works best for each family. In this article, shows that by including a physicochemical property together with the 3D information in a local structural element, fact improves the accuracy of remote homology detection. The ROC score for a set of models that includes the hydropathy index reaching a score of 0. 953 for the SCOP 1.53 data set. In addition, it is proposed an assembly model that uses the outputs obtained for the 10 properties to make a consensus decision. The consensus strategy achieves a ROC score of 0.963 on the SCOP 1.53 data set, surpassing the Current methods based on sequence composition whose accuracy ranges from 0.87 to 0.92.

• Autores: Bedoya Leiva, Oscar Fernando; Tischer, Irene
• Tipo de recurso: Article of journal
• Fecha de publicación: 2015
• Institución: Universidad del Valle
• Repositorio: Repositorio Digital Univalle
• Idioma: eng
• OAI Identifier: oai:bibliotecadigital.univalle.edu.co:10893/17747
• Acceso en línea: https://hdl.handle.net/10893/17747
• Palabra clave: Classification; Remote homology detection; 3D structural models physicochemical properties; SCOP family; Clasificación; Detección de homología remota; Modelos estructurales 3D propiedades fisicoquimicas; Familia SCOP; Clasificación; Detección de homología remota; Modelos estructurales 3D propiedades fisicoquímicas; Familia SCOP
• Rights: closedAccess
• License: http://purl.org/coar/access_right/c_14cb