Learning with multiple pairwise kernels for drug bioactivity prediction.

Journal: Bioinformatics (Oxford, England)
Published Date:

Abstract

MOTIVATION: Many inference problems in bioinformatics, including drug bioactivity prediction, can be formulated as pairwise learning problems, in which one is interested in making predictions for pairs of objects, e.g. drugs and their targets. Kernel-based approaches have emerged as powerful tools for solving problems of that kind, and especially multiple kernel learning (MKL) offers promising benefits as it enables integrating various types of complex biomedical information sources in the form of kernels, along with learning their importance for the prediction task. However, the immense size of pairwise kernel spaces remains a major bottleneck, making the existing MKL algorithms computationally infeasible even for small number of input pairs.

Authors

  • Anna Cichonska
    Department of Computer Science, Helsinki Institute for Information Technology HIIT, Aalto University, Espoo, Finland.
  • Tapio Pahikkala
    University of Turku, Turun Yliopisto, Turku, Finland. Electronic address: tapio.pahikkala@utu.fi.
  • Sandor Szedmak
    Department of Computer Science, Helsinki Institute for Information Technology HIIT, Aalto University, Espoo, Finland.
  • Heli Julkunen
    Department of Computer Science, Helsinki Institute for Information Technology HIIT, Aalto University, Espoo, Finland.
  • Antti Airola
    Department of Information Technology, University of Turku, Turku, Finland.
  • Markus Heinonen
    Department of Computer Science, Helsinki Institute for Information Technology HIIT, Aalto University, Espoo, Finland.
  • Tero Aittokallio
    Institute for Molecular Medicine Finland (FIMM), University of Helsinki, FI-00290 Helsinki, Finland.
  • Juho Rousu
    Department of Computer Science, Aalto University, 00076, Aalto, Finland. juho.rousu@aalto.fi.