arundurvasula / gal_bal

gal_bal

Pipeline for ABC simulations of multilocus balancing selection and introgression

1. Run bash init.sh to setup the directory stucture
2. Run qsub scripts/pipeline_introgression.sh to get the simulations for introgression

Note: make sure SLiM is available on your path. You can install it via the following:

# This script is for SLiM 3.2.1. Might not work for newer version of SLiM.
qrsh -l h_data=8G
cd ~
rm SLiM.zip

#download SLiM
wget http://benhaller.com/slim/SLiM.zip
unzip SLiM.zip

#make build directory
mkdir slim_build
cd slim_build

#load newer version of gcc
module load gcc/4.9.3

#tell cmake which one to use and compile
CC=gcc CXX=g++ cmake ../SLiM
make slim

Parameters for simulations

Introgression simulation

4N until MRCA of all S .cerevisae.

Ne = 10,000,000

migration rate - m uniform(0,1)

talpha = uniform(0,40,000,000)

Fixed parameters

Selfing rate = 1/50,000 (Kruglyak NG)

Clonning rate = 1/50,000 (use instead of selfing rate)

Gene conversion rate = (this is probably what yeast uses for the most part)

Recombination rate = 3.133483e-06 (BYxRM linkage mapping)

mutation rate = 3.8e-10

number of generations = 3.2 billion.

Notes:

To set the generation time as a variable, see here: https://groups.google.com/forum/#!topic/slim-discuss/0jOCti_t380

Use -t and -m to benchmark time and memory from the command line.

gal_bal's Issues

Make sure the between class diversity isn't just anchored with m2

Currently process_summary_stats.R anchors using m2 for the entire analysis, but if the eLD is less than 0.666667 then I need to make sure that for each locus it actually uses the between locus identity. This will probably just reduce the decay signal slightly and probably has no overall affect on the traces worth fixing for the next round of simulations.