Soft sweep detection is sensitive to assumptions about sweep prevalence

Stephen Rong, Sohini Ramachandran, Brown University

Questions: Contact stephen[underscore]rong[at]brown[dot]edu or post a git issue

Contents:

This git repo contains the following directories:

1. /scripts/simulations/ scripts for performing the simulations and calculating summary statistics

2. /scripts/applications/ scripts for preprocessing the 1000 Genomes Phase 1 raw haplotype data

3. /scripts/analyses/ scripts for generating temporary and final figures and results

4. /results/ used to contain the generated temporary and final figures and results

5. /data/applications/1000GP_Phase1_raw/ contains the 1000 Genomes Phase 1 raw haplotype data

6. /1000GP_Phase1_raw/ contains the preprocessed 1000 Genomes Phase 1 haplotype data

7. 1000GP_Phase1_misc/ contains extra files used in the analysis, explained in README.txt

8. 1000GP_Phase1_stan/ contains the SNP-level summary statistics used to standardize iHS, DiHH, and XP-EHH for the application data

9. 1000GP_Phase1_main/ contains the window-level summary statistics used as the application data

10. /data/simulations/2017111501/ contains the SNP-level summary statistics used to standardize iHS, DiHH, and XP-EHH for the simulation data

11. /data/simulations/2017111502/ contains the window-level summary statistics split into two disjoing sets for the training data and the test data

12. /data/simulations/2017111502/ contains the window-level summary statistics used to assess performance on neutral sits linked to a beneficial mutation

13. /hapbin_in_temp/ and /hapbin_out_temp/ are used for intermediate hapbin files

Software Requirements:

Python 2.7.13 -- numpy 1.11.3, pandas 0.19.2, pygg 0.1.7, networkx 1.11, matplotlib 2.0.0, scipy 0.18.1, sklearn 0.18.1

R 3.4.1 -- RColorBrewer 1.1-2, tidyverse 1.2.1, wrapr 1.2.0, ggthmr 1.1.0, mclust 5.4, superheat 0.1.0, ggbio 1.26.1, GenomicRanges 1.30.3, BSgenome.Hsapiens.UCSC.hg19 1.4.0

msms (https://github.com/delt0r/msms), hapbin (https://github.com/evotools/hapbin)

Generating Training, Test, and Application Data:

Download 1000 Genomes Phase 1 raw haplotypes from here (accessed on 12/8/2016), and unzip contents to /data/applictions/1000GP_Phase1_raw/. To generate contents of /data/applications/1000GP_Phase1_post/, go to /scripts/applications/ and run preprocess_data.py. Make sure modules in /scripts/simulations/ are in path and hapbin is installed.

To generate contents of /data/applications/1000GP_Phase1_stan/, go to /scripts/simulations/slurm_scripts/ and run scripts_apps_stan.py, and then run 1000GPP1_stan.sh from /scripts/simulations/. To generate contents of /data/simulations/2017111501/, go to /scripts/simulations/slurm_scripts/ and run scripts_sims_stan.py, and then run 2017111501.sh from /scripts/simulations/.

Then go to /scripts/analyses/ and run standardize_apps.R and standardize_sims.R.

To generate contents of /data/applications/1000GP_Phase1_main/, go to /scripts/simulations/slurm_scripts/ and run scripts_apps_main.py, and then run 1000GPP1_main.sh from /scripts/simulations/. To generate contents of /data/simulations/2017111502 and /data/simulations/2017111502, go to /scripts/simulations/slurm_scripts/ and run scripts_sims_train.py and scripts_sims_test.py, and then run 2017111502.sh and 2017111503.sh from /scripts/simulations/.

To generate result figures and tables, go to the /scripts/analyses/ directory and run master_script.R from within R, with the first two lines uncommented.