Code at https://github.com/josephryan/sowhat. Preprint at http://biorxiv.org/content/early/2015/05/07/005264
TEST 1 - CLASS DATA
module load bioware
cp -r /class/shared/SOWHAT_DEMO .
./sowhat --aln=class_data.phy --constraint=other_hyp.tre --raxml_model=GTRGAMMA --dir=class_data --name=class --reps=200
--aln = alignment file in PHYLIP sequential format
--constraint = alternative hypothesis to be tested
--raxml_model = model of evolution
--dir = output directory
--name = name of output files
--reps = maximum simulated samples to generate * this is optional in real analyses *
Results printed to class_data/sowhat.results
Results are continuously printed after 10 samples, p-values & confidence intervals are calculated after each sample
NOTE: If you need to rerun the analysis, delete the directory generated using --dir or attach the option --restart
What does this p-value indicate about our alternative hypothesis?
After how many samples did the conf. intervals show sufficient sampling?
TEST 2 - MORE DATA
./sowhat --aln=game_simul.phy --constraint=other_hyp.tre --raxml_model=GTRGAMMA --dir=game_sim --name=game --reps=200
results printed to game_sim/sowhat.results
how do the results of this test compare to the previous?
class_data.phy = sequences generated from our game - number of taxa = 4, number of sites = 27
game_simul.phy = longer sequences generated using game parameters - number of taxa = 4, number of sites = 1000
other_hyp.tre = tree incongruent with ML tree - internal node has been "flipped"