Spectral algorithms for heterogeneous biological networks

Martin Dean McDonald, Desmond Higham, J. Keith Vass

Research output: Contribution to journalArticlepeer-review


pectral methods, which use information relating to eigenvectors, singular vectors and generalized singular vectors, help us to visualize and summarize sets of pairwise interactions. In this work, we motivate and discuss the use of spectral methods by taking a matrix computation view and applying concepts from applied linear algebra. We show that this unified approach is sufficiently flexible to allow multiple sources of network information to be combined. We illustrate the methods on microarray data arising from a large population-based study in human adipose tissue, combined with related information concerning metabolic pathways.
Original languageEnglish
Pages (from-to)457-468
Number of pages12
JournalBriefings in functional genomics
Issue number6
Publication statusPublished - 2012


  • assortativity
  • eigenvector
  • Fiedler vector
  • Laplacian
  • meta-analysis
  • microarray
  • reordering
  • singular vector

Fingerprint Dive into the research topics of 'Spectral algorithms for heterogeneous biological networks'. Together they form a unique fingerprint.

Cite this