Iterative random forests to discover predictive and stable high-order interactions

Abstract

We developed a predictive, stable, and interpretable tool: the iterative random forest algorithm (iRF). iRF discovers high-order interactions among biomolecules with the same order of computational cost as random forests. We demonstrate the efficacy of iRF by finding known and promising interactions among biomolecules, of up to fifth and sixth order, in two data examples in transcriptional regulation and alternative splicing.

Document Details

Document Type
Pub Defense Publication
Publication Date
Jan 19, 2018
Source ID
10.1073/pnas.1711236115

Entities

People

  • Bin Yu
  • James B. Brown
  • Karl Kumbier
  • Sumanta Basu

Organizations

  • Army Research Office
  • Cornell University
  • National Human Genome Research Institute
  • National Science Foundation
  • Office of Naval Research
  • United States Department of Energy
  • United States National Library of Medicine
  • University of Birmingham

Tags

Readers

  • Computational Fluid Dynamics (CFD)
  • Molecular Genetics
  • Neural Network Machine Learning.