LDpop is a program for computing 2-locus likelihoods under the coalescent with recombination. Unlike previous methods, LDpop can correctly account for variable (but piecewise-constant) population size history.
LDpop produces lookup tables that can be used by other programs to estimate recombination maps. Other programs that can use the output of LDpop include:
LDpop also provides utilities for efficient posterior sampling of 2-locus ARGs.
Prerequisites:
- Python 2.7, 3.5, or 3.6
- Optional: Java 8
- Not required for computing lookup tables.
- Required for posterior sampling of 2-locus ARGs.
To install, in the top-level directory of LDpop (where "setup.py" lives), type
pip install .
Use run/ldtable.py to create a lookup table. See
run/ldtable.py --help
for usage.
By default run/ldtable.py uses an exact algorithm to compute the likelihoods.
To use a reasonable approximation that is much faster and scales to larger sample sizes,
use the flag --approx.
run/ldproposal.py and run/ImportanceSampler.jar are for importance sampling from the posterior distribution of 2-locus ARGs.
run/ldproposal.py creates a proposal distribution, that run/ImportanceSampler.jar uses to sample the ARGs. See their --help for instructions.
Also, see the examples.
Jack Kamm, Jeffrey Spence, Jeffrey Chan, Yun S. Song
Kamm, J.A.*, Spence, J.P.*, Chan, J., and Song, Y.S. Two-Locus Likelihoods under Variable Population Size and Fine-Scale Recombination Rate Estimation. Genetics, 2016. http://www.genetics.org/content/203/3/1381