Hands On 1: a Basic Rogue Taxon Search

This page provides step-by-step instructions on how to run RogueNaRok (RNR) and programs delivered with this package. It assumes, you correctly installed RoguNaRok.

Table of Contents Download dataset Basic analysis Pruning rogue taxa The reduced consensus tree

Download dataset

Download this test dataset and unzip it into your RNR folder. It contains 1000 bootstrap trees (*.bs) and a maximum likelihood estimate tree (*.tre) for an alignment with 150 taxa and an alignment with 2000 taxa. The files are in Newick format as output by RAxML.

Basic analysis

Without any options provided, RNR identifies rogues that (if removed) improve the support in a majority rule (MR) consensus tree. The basic call is:

    ./RogueNaRok -i data/150 -n runId  

Initially, RNR states (initScore = 0.570381, numBip=4183), that is, all in all there 4183 bipartitions in the tree set and the MR consensus tree has an RBIC of 57%. The file RogueNaRok_droppedRogues.runId contains the RogueNaRok-output of the rogue taxa search. We see that RNR could improve the RBIC of the consensus tree 61.7%.

Pruning rogue taxa

Now we want to prune the identified rogue taxa from the underlying bootstrap trees. rnr-prune takes a list of rogue taxa (one taxon per line) and a bootstrap tree collection and removes all taxa in the list from the trees. The following Linux command converts the RNR output into a list of rogue taxa:

    cat RogueNaRok_droppedRogues.runId  | cut -f 3  | tail -n  +3  | tr "," "\n"  > excludeThose.txt

The part before the first pipe (|) prints the file, then we extract the third column, remove the first two lines and convert multiple taxa per line into the format needed by rnr-prune. The output is directed to the file excludeThose.txt.

Maybe you do not want to prune all rogue taxa in the list, because some only lead to a very slight RBIC improvement. Add something like " | head -n 13" after the tail-command and you will only prune the 13 most deleterious rogue taxa. Note that, for a taxon R it is important that all taxa in the list up to R have been pruned and therefore we do not recommend to prune a random selection of taxa in the list (except you want to explore the "pruning space" on your own). If you want to avoid that RogueNaRok prunes a certain set of taxa, conduct a search using the exclude option.

Finally, we can prune the taxa in the file with

    ./rnr-prune  -i data/150.bs  -x excludeThose.txt  -n reduced 

Pruned bootstrap trees are written to RnR-prune_prunedBootstraps.reduced.

The reduced consensus tree

Finally, for the reduced consensus tree, invoke RAxML:

     /path/to/raxmlHPC-SSE3 -z RnR-prune_prunedBootstraps.reduced  -m GTRCAT -J MR -n tmp

It is easy to check the resolution of this consensus tree:

    cat RAxML_MajorityRuleConsensusTree.tmp  | tr "[" "\n"  | tr "]"  "\n" | grep "^[0-9]*$" | wc -l 

or the sum of the support:

    cat RAxML_MajorityRuleConsensusTree.tmp  | tr "[" "\n"  | tr "]"  "\n" | grep "^[0-9]*$" | awk '{TMP+=$1} END {print TMP/100}'

.