Given an undirected weighted graph, in which each vertex is assigned to a color and one of them is identified as source, in the all-colors shortest path problem we look for a minimum cost shortest path that starts from the source and spans all different colors. The problem is known to be NP-Hard and hard to approximate. In this work we propose a variant of the problem in which the source is unspecified and show the two problems to be computationally equivalent. Furthermore, we propose a mathematical formulation, a compact representation for feasible solutions and a VNS metaheuristic that is based on it. Computational results show the effectiveness of the proposed approach for the two problems.