This repository contains information for reproducing experiments in the HaploADMIXTURE manuscript:
To be updated
To download the 1000 Genomes Project (TGP), Human Genome Diversity Project (HGDP), and Human Origins (HO) public data sets, follow the steps in https://github.com/sriramlab/SCOPE/tree/master/misc/real_data and https://github.com/StoreyLab/terastructure/tree/master/scripts/data.
The files Project.toml and Manifest.toml define Julia package environment for the experiments. To install all the packages,
- launch Julia with
--project=<path-to-this directory>or activate the environment withusing Pkg; Pkg.activate(normpath("<path-to-this directory>")), - then run
using Pkg; Pkg.instantiate().
Please note that the name of the package for sparse K-means via featuer ranking has changed from SKFR to SparseKmeansFeatureRanking. This repository will be updated once SparseKmeansFeatureRanking, OpenADMIXTURE, and HaploADMIXTURE are all registered.
Each real data directory (TGP, HGDP, HO) contains the following:
filter_[].jl: filtering the original datasets using SKFR.run_CV.jl: (4-fold) cross-validation for selecting the value ofK, the number of populations, the training step.run_CV_test.jl: (4-fold) cross-validation for selecting the value ofK, the number of populations, the test step.run_[]_Ks.jl: Run the HaploADMIXTURE model for different values ofK, to selectKusing Akaike information criterion.run_[]_adm.jl: Run the original admixture model using OpenADMIXTURE.run_[]_full.jl: Run HaploADMIXTURE with all the SNPs in the original dataset.run_[].jl: Run HaploADMIXTURE with the dataset filtered withfilter_[].jl.run_admix_full.jl: Run OpenADMIXTURE with the original dataset.
Please note that the released version of the package on https://github.com/OpenMendel/HaploADMIXTURE.jl contains the function run_admixture() to simplify the code.
For the Simulation directory, there is the simulation.jl file to generate datasets for simulation.