Tally Approximations of Phylogenetic Informativeness Rapidly (TAPIR)
tapir contains programs to estimate and plot phylogenetic informativeness for large datasets.
When using tapir, please cite:
- Faircloth BC, Chang J, Alfaro ME: tapir enables high throughput analysis of phylogenetic informativeness. https://github.com/faircloth-lab/tapir
- Townsend JP: Profiling phylogenetic informativeness. Systematic Biol. 2007, 56:222-231.
- Pond SLK, Frost SDW, Muse SV: HyPhy: hypothesis testing using phylogenies. Bioinformatics 2005, 21:676-679.
- hyphy2 (please download or build a single-threaded hyphy2)
- Python 2.6
For ALL platforms, you must download a hyphy binary for your platform (osx or linux) and place that within your $PATH:
wget https://github.com/downloads/faircloth-lab/tapir/hyphy2.osx.gz gunzip hyphy2.*.gz chmod 0700 hyphy2.* mv hyphy2.* ~/Bin/hyphy2
To install the other dependencies (numpy, scipy), you may need to install a Fortran compiler on linux/osx:
On linux (ubuntu/debian), use:
apt-get install gfortran libatlas-base-dev liblapack-dev
Install tapir and dependencies, which include numpy and scipy (the reason we installed the dependencies above):
pip install tapir
To plot results, you will also need to:
apt-get install r-base r-base-dev pip install rpy2
It is easiest just to install the scipy superpack. This will install the dependencies that tapir needs. After installing the superpack, using pip, install tapir:
pip install tapir
Alternatively, you can simply try to install tapir using:
pip install tapir
To plot results, you need to install R and then install rpy2:
pip install rpy2
wget http://pypi.python.org/packages/source/t/tapir/tapir-1.0.tar.gz tar -xzvf tapir-1.0.tar.gz cd tapir* python setup.py build python setup.py test python setup.py install
Plotting is optional. To install the plotting dependencies, see Installation, above.
If you didn’t run the tests using
python setup.py test above, you can also:
import tapir tapir.test()
estimate_p_i.py code calls a batch file for hyphy that is in
templates/. This file needs to be in the same position relative to
wherever you put
estimate_p_i.py. If you install thins as above, you’ll
be fine, for the moment.
cd /path/to/tapir/ python tapir_compute.py Input_Folder_of_Nexus_Files/ Input.tree \ --output Output_Directory \ --epochs=32-42,88-98,95-105,164-174 \ --times=37,93,100,170 \ --multiprocessing
--multiprocessing is optional, without it, each locus will be run
If you have already run the above and saved results to your output folder (see below), you can use the pre-existing site-rate records rather than estimating those again with:
python tapir_compute.py Input_Folder_of_Site_Rate_JSON_Files/ Input.tree \ --output Output_Directory \ --epochs=32-42,88-98,95-105,164-174 \ --times=37,93,100,170 \ --multiprocessing \ --site-rates
You can access the results in the database as follows. For more examples, including plotting, see the documentation
crank up sqlite:
get integral data for all epochs:
select locus, interval, pi from loci, interval where loci.id = interval.id
get integral data for a specific epoch:
select locus, interval, pi from loci, interval where interval = '95-105' and loci.id = interval.id;
get the count of loci having max(PI) at different epochs:
create temporary table max as select id, max(pi) as max from interval group by id; create temporary table t as select interval.id, interval, max from interval, max where interval.pi = max.max; select interval, count(*) from t group by interval;
tapir contains plotting scripts to help you plot data within a results database and compare data between different databases. tapir uses RPY and R to do this. You can also plot data directly in R. Until we finish the documentation, please see the wiki for examples.
BCF thanks SP Hubbell, PA Gowaty, RT Brumfield, TC Glenn, NG Crawford, JE McCormack, and M Reasel. JHLC and MEA thank J Eastman and J Brown for thoughtful comments about PI. We thank Francesc Lopez-Giraldez and Jeffrey Townsend for providing us with a copy of their web-application source code and helpful discussion.