Supplementary Data: Ufboot2: Improving The Ultrafast Bootstrap Approximation

Supplementary Data<br> UFBoot2: Improving the Ultrafast Bootstrap Approximation<br> doi: https://doi.org/10.1101/153916<br> http://www.biorxiv.org/content/early/2017/06/22/153916 This record contains PANDIT based dataset and TreeBASE dataset (Nguyen et al. 2015) which are analyzed by different bootstrap methods in the study "UFBoot2: Improving the Ultrafast Bootstrap Approximation". The PANDIT based dataset (compressed in file data_pandit.tar.gz) is used to benchmark the accuracy of bootstrap estimates. The TreeBASE dataset (compressed in file data_treebase.tar.gz) is used to benchmark runtimes. After being uncompressed, the PANDIT based dataset comprises: 5,690 numbered directories corresponding to 5,690 DNA MSAs simulated by Seq-Gen (Rambaut and Grass 1997), where the model parameters and true tree were inferred from the original MSAs downloaded from the PANDIT database (Whelan et al. 2006). Note that the numbering of these directories is not consecutive because we kept only MSAs that can be tested under the mild and severe model violations as defined in the UFBoot paper (Minh et al. 2013). In each numbered directory N, there are three files: (1) data.N contains the simulated MSA in PHYLIP format; (2) model.N contains the best-fit model detected from the corresponding original MSA; (3) tree.N contains the tree (in Newick format) inferred from the corresponding original MSA. tree.N and model.N are used by Seq-Gen to simulate the MSA in data.N. After being uncompressed, the TreeBASE dataset comprises 115 files corresponding to 115 MSAs. There are: 70 DNA MSAs in PHYLIP format. These files follow the naming scheme dna_[number of sequences]_[number of sites].phy. 45 protein MSAs in PHYLIP format. These files follow the naming scheme prot_[number of sequences]_[number of sites].phy.