The police here, known for their summary methods, say they have already arrested around 100 people in connection with Sunday's killings.
Therefore, alternative summary methods should be explored.
Here, we point out some of the classes with high and low correspondence in each case using three summary methods: biplots from the first three dimensions of each correspondence analysis (Figs. 2 and 3), heat maps of species similarity (Fig. 4), and the full contingency tables of cross-tabulated plot frequencies (Additional file 3).
However, all summary methods we explored were impacted by gene tree estimation error.
ASTRAL was generally more accurate than all the other summary methods (Fig. 2).
However, even the fast summary methods can be computationally intensive on large datasets.
Thus, despite the theoretical advantages of coalescent-based summary methods, there is a heated debate about whether summary methods or a maximum likelihood concatenated analysis would more accurately estimate phylogenies under biologically realistic conditions [ 23, 24, 2, 28, 29].
However, for a few cases, ASTRAL and one or more summary methods had similar accuracy; for example, on 800 true gene trees from default ILS levels, all summary methods (except for Greedy) produced the true species tree.
Finally, while ASTRAL-2 is the most accurate of the summary methods, the difference between ASTRAL-2 and the other summary methods is small (ranging from 0.3% to 1.9%).
Since summary methods are able to analyze large datasets (some are sufficiently fast that they can analyze datasets with 1000 species and 1000 genes [ 6]), most coalescent-based analyses of biological datasets have been performed using summary methods [ 18, 19].
