On the accuracy of metadynamics and its variations in a protein folding process