This page contains the supplementary data for the manuscript:

Bui Quang Minh, Matthew W. Hahn, and Robert Lanfear. New methods to calculate concordance factors for phylogenomic datasets. Submitted.

A tutorial how to perform concordance factor analysis is available here: http://www.iqtree.org/doc/Concordance-Factor.

Files in this repository

• Rodriguez_2018.tar.gz: the Lizard data set containing 3220 UCE alignments kindly provided by Zachary Rodriguez (https://doi.org/10.1126/sciadv.aao5017). It contains the alignment.nex file in NEXUS format for the concatenated alignment with partitioning information.
• outputs/partitions.spp.treefile.rename: Newick tree used to draw Figure 2.
• scripts/: This folder contains the bash scripts that we used to run IQ-TREE under Linux (see below for the detailed command lines) and R scripts to make Figure 3 (figure3.R) and the supplementary figures (make_supplementary_figures.R).

Lizard dataset

In order to replicate the analysis for the Lizard dataset (Rodriguez et al. 2018), please do the following:

1. Unzip the Lizard data file into a folder Rodriguez_2018:

    tar -xzf Rodriguez_2018.tar.gz
   

2. Go inside this folder:

    cd Rodriguez_2018
   

3. Infer a species tree using partition model and ultrafast bootstrap (assuming that iqtree command points to the IQ-TREE version 2.0-rc1):

    iqtree -p alignment.nex --prefix concat -bb 10000 -nt AUTO -seed 609149
   

4. Infer separate locus trees (one for each UCE):

    iqtree -S alignment.nex --prefix loci -nt 10 -seed 937413
   

5. Compute gene and site concordance factors:

    iqtree -t concat.treefile --gcf loci.treefile -p alignment.nex --scf 100 --prefix concord
   

The resulting newick tree concord.cf.tree will have branches annotated with bootstrap proportion, gene and site concordance factors separated by slash (/). This tree file can be viewed by e.g. FigTree. Note that the sCF values might be slightly different from the output newick tree provided here due to random sampling of the quartets.

Detailed tree branch statistics are printed to concord.cf.stat and tree with branch ID concord.cf.branch.

Subsampling analysis

The other nine datasets (Table 1 of the main text) are obtained from https://github.com/roblanf/BenchmarkAlignments, with the link to FigShare to download the alignments: https://figshare.com/s/622e9e0a156e5233944b. Unzipping the file will create a directory for each dataset (e.g., Ballesteros_2019), which contains an alignment.nex file for the alignment in NEXUS format. You should remove the CHARSET nuclear_genome = ... line in this file.

To analyze a subsampled alignment of 10, 20, ..., 200 loci, go to the directory for each dataset and run the ultrafast boostrap (UFBoot), single-locus tree inference and standard boostrap (StdBoot) with a for loop in Linux:

threads=10 # can be changed depending on the machine
for ((k=10; k <= 200; k+=10)); do
	# perform ultrafast bootstrap (UFBoot)
	iqtree -p alignment.nex --prefix ufboot.$k -T $threads -m TEST -bb 1000 --no-terrace --subsample $k --subsample-seed 1

	# perform single-locus tree inference
  	iqtree -S alignment.nex --prefix loci.$k -T $threads -m TEST --subsample $k --subsample-seed 1

	# perform standard bootstrap, only for Rodriguez dataset
	iqtree -p alignment.nex --prefix stdboot.$k -T $threads -m TEST -b 100 --no-terrace --subsample $k --subsample-seed 1
done

Once done, we can now compute gCF and sCF onto the best 200-loci tree from the UFBoot outputs:

for ((k=10; k <= 200; k+=10)); do
	# map the UFBoot supports onto the 200-loci tree
	iqtree -sup ufboot.200.treefile -t ufboot.$k.splits.nex --prefix ufboot_sup.$k

	# further map the gCF and sCF onto this tree
	iqtree -t ufboot_sup.$k.suptree --gcf loci.$k.treefile -p alignment.nex --subsample $k --subsample-seed 1 --scf 100 --prefix s$k
done

Additionally for the Rodriguez_2018 dataset we do the same from the IQ-TREE outputs of the standard bootstrap:

for ((k=10; k <= 200; k+=10)); do
	# map the standard bootstrap supports onto the 200-loci tree
	iqtree -sup ufboot.200.treefile -t stdboot.$k.splits.nex --prefix stdboot_sup.$k

	# further map the gCF and sCF onto this tree
	iqtree -t stdboot_sup.$k.suptree --gcf loci.$k.treefile -p alignment.nex --subsample $k --subsample-seed 1 --scf 100 --prefix r$k
done

Reproducing the Figures

After completing the analyses above, the output of the analyses of all 10 datasets should be contained in a single folder, with one sub-folder for each dataset (one for the lizard dataset, a further 9 for the other 9 datasets). Each dataset folder should be named for the author and year, and in some cases the data type, as follows:

• Ballesteros_2019
• Branstetter_2017
• Cannon_2016
• Jarvis_2015
• Misof_2014
• Ran_2018_aa
• Ran_2018_dna
• Rodriguez_2018
• Wu_2018_aa
• Wu_2018_dna

Once this is done, you can recreate the figures as follows.

Figure 2 in the manuscript was drawn partly in FigTree and then edited in Adobe Illustrator, so it's not possible to reproduce it with a script.

To reproduce Figure 3, use the R script figure3.R[https://github.com/bqminh/concordance-factor/blob/master/scripts/figure3.R]. You will need to re-name the base variable to point to the parent folder of the 10 dataset folders you created above.

To reproduce the supplementary figures, use the R script make_supplementary_figures.R[https://github.com/bqminh/concordance-factor/blob/master/scripts/make_supplementary_figures.R]. You will need to re-name the base variable to point to the parent folder of the 10 dataset folders you created above.