1. Tandy Warnow The University of Illinois
New Scalable Coalescent-Based Species Tree Estimation Methods: BBCA, ASTRAL, and ASTRID
Tandy Warnow
The University of Illinois

2. BBCA, ASTRAL, and ASTRID
BBCA is a simple way of making *BEAST scalable to large numbers of genes (but doesn’t address large numbers of species)
ASTRAL and ASTRID:
summary methods that are statistically consistent in the presence of ILS, and that run in polynomial time.
Both can analyze very large datasets (1000 species and 1000 genes – or more) with high accuracy.
The relative accuracy depends on the model condition – sometimes ASTRAL is better, sometimes ASTRID is better.

4. Main competing approaches
gene gene gene k
. . .
Species
Concatenation
. . .
Analyze
separately
point out that supertree methods take overlaping trees and produce a tree, and that the whole process of first generating small trees and then applying a supertree method is often referred to as the “supertree approach”.
Summary Method

5. Incomplete Lineage Sorting (ILS) is a dominant cause of
gene tree heterogeneity

6. *BEAST Heled and Drummond, MBE 2010
Input: set of multiple sequence alignments for collection of genes
Techique: Uses MCMC to co-estimate gene trees and species trees
Highly accurate
Limited in practice to small numbers of genes and species, due to convergence issues

7. BBCA: improving *BEAST
Zimmermann, Mirarab, and Warnow, BMC Genomics 2014:
Randomly partition genes into bins of at most 25 genes
Apply *BEAST to each bin, and take the gene trees it computes
Apply favored summary method to the gene trees
Matches accuracy of *BEAST
Improves scalability to large # genes

8. ASTRAL Mirarab and Warnow, Bioinformatics 2014
Tutorial in Species Tree Workshop

11. ASTRID
ASTRID: Accurate species trees using internode distances, Vachaspati and Warnow, RECOMB-CG 2015 and BMC Genomics 2015
Algorithmic design: Computes a matrix of average leaf-to-leaf topological distances, and then computes a tree using FastME (more accurate than neighbor Joining and faster, too).
Related to NJst (Liu and Yu, 2010), which computes the same matrix but then computes the tree using neighbor joining (NJ).
Statistically consistent under the MSC
O(kn2 + n3) time where there are k gene trees and n species

14. BBCA, ASTRAL, and ASTRID
BBCA is a simple way of making *BEAST scalable to large numbers of genes (but doesn’t address large numbers of species)
ASTRAL and ASTRID:
summary methods that are statistically consistent in the presence of ILS, and that run in polynomial time.
Both can analyze very large datasets (1000 species and 1000 genes – or more) with high accuracy.
The relative accuracy depends on the model condition – sometimes ASTRAL is better, sometimes ASTRID is better.

15. Acknowledgments Software
ASTRAL: Available at
ASTRID: Available at
Others at
NSF grant DBI (joint with Noah Rosenberg at Stanford and Luay Nakhleh at Rice):
NSF graduate fellowship to Pranjal Vachaspati
HHMI graduate fellowship to Siavash Mirarab
Papers available at