Stochastic coordinate descent Frank-Wolfe algorithm for large-scale biological network alignment

With increasingly "big" data available in biomédical research, deriving accurate and reproducible biology knowledge from such big data imposes enormous computational challenges. In this paper, we propose a highly scalable randomized coordinate descent Frank-Wolfe algorithm for convex optimization with compact convex constraints, which has diverse applications in analyzing biomédical data for better understanding cellular and disease mechanisms. We focus on implementing the derived stochastic coordinate descent algorithm to align protein-protein interaction networks for identifying conserved functional pathways based on IsoRank. The stochastic algorithm naturally leads to the decreased computational cost for each iteration. More importantly, we show that it achieves a linear convergence rate. Our numerical test confirms the improved efficiency of this technique for the large-scale biological network alignment problem.