Fork me on GitHub

BOOSTER is an alternative method for calculating the branch supports of phylogenetic trees, using the bootstrap. With large trees and moderate phylogenetic signal, BOOSTER tends to be more informative than the standard Felsenstein bootstrap method.

BOOSTER's approach has a simple but sound statistical foundation. It replaces the branch presence proportion (i.e. the expectation of a {0,1} indicator function) of Felsenstein's bootstrap, by the expectation of a refined, gradual function in the [0,1] range, quantifying branch presence in the bootstrap trees.

BOOSTER is especially relevant with deep branches and large datasets, where branches known to be (nearly) correct are supported by the transfer bootstrap while they are not by Felsenstein's method. Though higher, BOOSTER supports are not impeded with falsely supported branches.

You need two files to run BOOSTER: one with the reference tree inferred using your preferred method; the second with the bootstrap trees inferred using the same method from the bootstrapped alignments. With multiple alignments of moderate size (< 3,000 taxa, < 10,000 sites), our webserver can also run PhyML and FastTree to infer the reference and bootstrap trees, and then run BOOSTER.

BOOSTER C code is open source and available on GitHub.

The transfer bootstrap is available in several phylogenetic programs: PhyML, SeaView, RaXML-NG, and others.

Please send your comments and/or suggestions to: frederic[dot]lemoine[at]pasteur[dot]fr and olivier[dot]gascuel[at]pasteur[dot]fr, Unité Bioinformatique Évolutive, C3BI USR 3756, Institut Pasteur and CNRS, Paris, France

News

  • 2017/11: PhyML and FastTree are available to infer trees from MSAs.
  • 2017/10: PhyML now implements Transfer Bootstrap (option --tbe, see GitHub repository)

More information

  • To learn more about BOOSTER, please read our help page;
  • The preprint with all details and results is available on bioRxiv.

References

  • Renewing Felsenstein's Phylogenetic Bootstrap in the Era of Big Data, Nature 556, 452-456 (2018)

    F. Lemoine, J.-B. Domelevo-Entfellner, E. Wilkinson, D. Correia, M. Davila Felipe, T. De Oliveira, O. Gascuel

    Access the recommendation on F1000Prime

Data

  • The archive containing all data described in the paper can be downloaded from the github page

Run BOOSTER

Run BOOSTER
twitter facebook linkedin youtube3