A graph-based algorithm for the multi-objective optimization of gene regulatory networks

TL;DR

This paper introduces a graph-based algorithm that efficiently computes Pareto optimal solutions for multi-objective gene regulatory network optimization problems, leveraging polytope mappings and dynamic programming.

Contribution

It presents a novel polynomial-time method for exhaustively finding Pareto optimal solutions in linear MOPs using coloured Hasse diagrams and edge contractions.

Findings

Exact Pareto sets can be computed in polynomial time.

Special cases allow for linear-time solutions.

The approach is applicable to complex gene regulatory network optimization.

Abstract

The evolution of gene regulatory networks in variable environments poses Multi-objective Optimization Problem (MOP), where the expression levels of genes must be tuned to meet the demands of each environment. When formalized in the context of monotone systems, this problem falls into a sub-class of linear MOPs. Here, the constraints are partial orders and the objectives consist of either the minimization or maximization of single variables, but their number can be very large. To efficiently and exhaustively find Pareto optimal solutions, we introduce a mapping between coloured Hasse diagrams and polytopes associated with an ideal point. A dynamic program based on edge contractions yields an exact closed-form description of the Pareto optimal set, in polynomial time of the number of objectives relative to the number of faces of the Pareto front. We additionnally discuss the special case of series-parallel graphs with monochromatic connected components of bounded size, for which the running time and the representation of solutions can in principle be linear in the number of objectives.