Statistics in Biosciences

, Volume 4, Issue 1, pp 3-26

First online:

An Efficient Optimization Algorithm for Structured Sparse CCA, with Applications to eQTL Mapping

  • Xi ChenAffiliated withMachine Learning Department, Carnegie Mellon University Email author 
  • , Han LiuAffiliated withBiostatistics Department, Computer Science Department, Johns Hopkins University

Rent the article at a discount

Rent now

* Final gross prices may vary according to local VAT.

Get Access


In this paper we develop an efficient optimization algorithm for solving canonical correlation analysis (CCA) with complex structured-sparsity-inducing penalties, including overlapping-group-lasso penalty and network-based fusion penalty. We apply the proposed algorithm to an important genome-wide association study problem, eQTL mapping. We show that, with the efficient optimization algorithm, one can easily incorporate rich structural information among genes into the sparse CCA framework, which improves the interpretability of the results obtained. Our optimization algorithm is based on a general excessive gap optimization framework and can scale up to millions of variables. We demonstrate the effectiveness of our algorithm on both simulated and real eQTL datasets.


Sparse CCA Structured sparsity Group structure Network structure Genome-wide association study eQTL mapping Optimization algorithm