Gravar-mail: Enhancing the accuracy of HMM-based conserved pathway prediction using global correspondence scores