Fig. 10.

Homology coverage benchmark: performance of the optimal-branch e2msa.AIF method both for global and local homologies. The reason why the PPV for the Global Homology set is not 100 % for in all case, is because the structural alignments from the benchmark PREFAB include some positions in lowercase which are not supposed to be considered aligned, thus there is the option of some small overextension into those positions. The results for the short-branch and long-branch parameterizations for the Local homology set are the same as those provided in Fig. 8 b now placed in reference to the optimal-branch parameterization