Graph data structures are commonly used in computer science for modelling real objects in a large number of applications. For instance, in bioinformatics and computational biology they are used to model proteins and their interactions in a living organism. In such a scenario nodes model proteins while edges represent interactions among them. Thus, graph based pattern recognition algorithms may be used to extract biologically meaningful knowledge from data. Such algorithms are in general computationally expensive. Conversely, vector-based representation are semantically less-powerful but enable the use of a large class of efficient algorithms. A possible solution for the trade-off among semantic expressiveness and computational tractability of representation is realized by a two step process: (i) the transformation of graph structures into vectors, (ii) the use of vector based pattern recognition algorithms. Despite this possibility, the use of such a workflow for protein interaction data remains still an unexplored field. This paper presents a methodology for the embedding of graphs representing protein interaction network into vectors and the subsequent analysis. The presented methodology is implemented into an available tool, VeNEt, showing the effectiveness of such an approach in a case study. VeNEt is easily customisable so it may be used in a large class of problems in biological networks

VeNet: A framework for the analysis of protein interaction networks through vector space embedding

Veltri Pierangelo;Cannataro M;Guzzi P
2011-01-01

Abstract

Graph data structures are commonly used in computer science for modelling real objects in a large number of applications. For instance, in bioinformatics and computational biology they are used to model proteins and their interactions in a living organism. In such a scenario nodes model proteins while edges represent interactions among them. Thus, graph based pattern recognition algorithms may be used to extract biologically meaningful knowledge from data. Such algorithms are in general computationally expensive. Conversely, vector-based representation are semantically less-powerful but enable the use of a large class of efficient algorithms. A possible solution for the trade-off among semantic expressiveness and computational tractability of representation is realized by a two step process: (i) the transformation of graph structures into vectors, (ii) the use of vector based pattern recognition algorithms. Despite this possibility, the use of such a workflow for protein interaction data remains still an unexplored field. This paper presents a methodology for the embedding of graphs representing protein interaction network into vectors and the subsequent analysis. The presented methodology is implemented into an available tool, VeNEt, showing the effectiveness of such an approach in a case study. VeNEt is easily customisable so it may be used in a large class of problems in biological networks
2011
978-145771613-3
Pattern Recognition
Protein Interaction Networks
Vector Space Embedding
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.11770/362546
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 1
  • ???jsp.display-item.citation.isi??? 1
social impact