If you only need the GSM extraction script (Scripts/Extract_GSM.ipynb), you only need to install the FaST-LMM part. If you want to use the entire GWAS pipeline, R as described below will also be required.

1. You need a Unix environment. (WSL (Windows Subsystem for Linux) works too.)
2. Install an Anaconda 2 for Linux distribution: https://www.anaconda.com/download/#linux
3. This pipeline uses a modified version of FaST-LMM from Microsoft Genomics. Once Anaconda python is installed, download the full GWAS_Pipeline project, and in its FaST-LMM folder type sudo python setup.py install. After this, FaST-LMM is fully functional.
   Important: Make sure to install FaST-LMM from the folder within this pipeline, and NOT the one from the Microsoft Genomics Github page.
4. The Manhattan plot provided by Microsoft Genomics doesn't always work very well, therefore another one is used. Install it by typing pip install https://github.com/khramts/assocplots/archive/master.zip

Warning: FaST-LMM was updated only recently. To make sure to have the latest pysnptools, just type the following in your Anaconda command prompt:
pip uninstall pysnptools
pip install pysnptool

Install any R distribution in your Unix environment. Then type

sudo apt install libssl-dev        # openssl compatibility
sudo apt install libxml2-dev       # "
sudo apt install gfortran          # "  

This (should) make(s) sure that Unix R runs the phenotype adjustments without errors.

Run the following code in the project's top-level folder:

chmod +x Bash_GWAS_Pipeline_Full.sh
chmod +x PipelinePart2_Plink2FilteringAlleles.sh
chmod +x plink2_linux_x86_64/plink2

This pipeline

1. adjusts your raw individual phenotype measurements for inversions and Wolbachia infections

2. Starting with the variants present in the original DGRP2 variant files (.bed, .bim, .fam triad) from the official DGRP2 website, filters to only keep variants with MAF>WhateverThresholdYouChoose (default is 0.05). The number of kept variants will be specific to your phenotype, i.e. depending on which lines where used in the specific experiments

3. performs single SNP GWAS on the adjusted phenotype and filtered variants, and stores the results as a file in the Outputs/ directory.

   • This pipeline can do permutations.

• dgrp2.bed, *.bim and *.fam, freeze2.common.rel.mat, wolbachia.csv, inversion.csv (all from the official DGRP2 site, wolbachia and inversion have to be converted from .xlsx to .csv) to Raw_Data/
• Your phenotype files to Inputs/. They must be named *Phenotype_Name*_Phenotype_Full.txt

1. Put your phenotype in the Inputs/ folder and name it *Pheno_Name*_Phenotype_Full.txt. It should follow the formatting line_id \t phenotype value \t sex (m/f), and should not have a header.

2. In the unix environment of your choice, in the main folder of the project, type
   var1=value1 var2=value2 ... ./Bash_GWAS_Pipeline_Full.sh\
   
   Any of the variables are optional and can be omitted! The variables that you can choose from, are:

Example: pheno=Mass sex=Male perm=5 reproducibility=FALSE ./Bash_GWAS_Pipeline_Full.sh

Open Scripts/Extract_GSM.ipynb in Jupyter. More information is given in the notebook itself.