Quick run

The basic things that you need to run treePL are 1) a tree file that is just a simple file with a newick string per line and 2) a configuration file.

The configuration file, in its most basic form would look like:

treefile = intree.tre
smooth = 100
numsites = 8502
mrca = EUDICOT Papaver_nudicaule Lonicera_etrusca
min = EUDICOT 123
max = EUDICOT 140
outfile = intree.dated.tre

The mrca should be in the form mrca = NAME taxon1 taxon2 ... with the constraints being min = MRCANAME date and max = MRCANAME date. To fix an age just give the same min and max. Comments can just be prefixed with a # or [ (or really anything that isn't a letter--treePL will just ignore it).

Two additional commands can be very helpful. One is thorough and the other is prime. For thorough, it will make sure that it continues to iterate until convergence. It can take longer but is probably a good idea. Just add thorough to a line like this:

treefile = intree.tre
smooth = 100
numsites = 8502
mrca = EUDICOT Papaver_nudicaule Lonicera_etrusca
min = EUDICOT 123
max = EUDICOT 140
outfile = intree.dated.tre
thorough

For prime, you would only run this once and it will test different optimization possibilities. At the end of the run it will let you know which performed the best (you can also see for yourself). Then it will give you some options to put at the bottom of you configuration file. To use this just add prime to a line in the configuration file like this:

treefile = intree.tre
smooth = 100
numsites = 8502
mrca = EUDICOT Papaver_nudicaule Lonicera_etrusca
min = EUDICOT 123
max = EUDICOT 140
outfile = intree.dated.tre
thorough
prime

You will want to comment this out after running once.

To run cross validation, you should add

cv

to the configuration file. You can also add

randomcv

to run it with the RSRCV cross validation. The output will be in cv.out or a different file if you designate

cvoutfile = filename

in the configuration file. treePL will use the best chisq for the final smoothing, but you should probably replicate and check the stability of results. Then you can choose the best smoothing.

To run with multiple threads set

nthreads = number

.