Tutorial to run GWAS including the X chromosome

• This tutorial will cover the following steps, all including the autosomes and the X chromosome:
  • Quality control of genotype data
  • Imputation of genotype data
  • Genome-wide association study (GWAS)
• Genotype data is provided to use with this tutorial, or you can run through it using your own genotype data

Getting Started

Clone the repository

• Clone this repository to get a local copy of all files

• In your terminal or HPC environment navigate to the location you would like to download a copy of all the files in this tutorial

• Then run the following code:

git clone https://github.com/jodithea/Tutorial_GWAS_including_X_chromosome.git

• You will now have a copy of all of the directories and files from this tutorial

Software needed

• For this tutorial you will need the following software
  • Plink v1.9
  • R (v4.5.0 is used in this tutorial)
    • Make sure the R package 'tidyverse' is installed
  • vcftools (v0.1.16 used in this tutorial)
  • bcftools (v1.22 used in this tutorial)
  • GCTA (v1.94.1 used in this tutorial)

How to use this tutorial

• Follow through the directories and files in chronological order

Markdown files

• Markdown files (files ending with .md) contain instructions and documentation
• You can view these files directly in the GitHub web interface
• Or, if you have a local clone of the repository, you can view them in the terminal using cat, e.g.:

 cat README.md

Scripts

• Scripts (i.e. files ending with '.sh') are shell scripts that can be submitted to your HPC environment
• Again you can view these files directly in the GitHub web interface or in your local clone of the repository by using cat
• All scripts in this repository follow a similar layout:

1. Shebang and Scheduler Header
   • Every script starts with a shebang (#!/bin/bash) to indicate it is a bash script, followed by HPC scheduler directives
   • The schedular directives are written for PBS, for example:

#!/bin/bash
#PBS -l walltime=00:10:00
#PBS -l mem=1GB
#PBS -l ncpus=1

If you use SLURM instead of PBS, you can replace these lines with SLURM directives, for example:

#!/bin/bash
#SBATCH --time=00:10:00
#SBATCH --mem=1GB
#SBATCH --cpus-per-task=1

To do this replacement, in your local clone of the repository, open the script with an editor, for example:

nano 01_Split_X_chromosome.sh

Then edit the scheduler directives in the file as needed

2. Script Description

   • A short comment explaining what the script does, for example:

# This script splits the X chromosome into non-pseudoautosomal region (nPAR) (coded as CHR 23) and pseudoautosomal region (PAR) (coded as CHR 25)

3. Environment

   • This section loads any required modules or software, for example:

### Environment ###
module load plink/1.90b7

4. Preamble / User Variables

   • This section defines any directories, file paths, or parameters used in the script
   • You need to edit this section of the script as appropriate
   • For example, in each script the 'directory' variable is defined. Open the script using 'nano' (or a similar editor) and set the path to the location of where your local clone of the repository is

### Preamble ###
# Update to point to the location where you are doing this tutorial
directory=/path/Tutorial_GWAS_including_X_chromosome/

5. Commands / Submission Section

   • This section contains the actual commands that perform the work
   • For example:

### Submit script ###
cd ${directory}

# Split chromosome X
plink --bfile 01_Genotype_Data/Genotypes_chrX \
      --split-x b37 \
      --make-bed \
      --out 01_Genotype_Data/Genotypes_chrX_split

Submitting scripts

• After viewing the script and making any edits as appropriate, submit the script

For PBS, submit using qsub, e.g.:

qsub 01_Split_X_chromosome.sh

For SLURM (and after you have edited the scheduler directives appropriately), submit using sbatch, e.g.:

sbatch 01_Split_X_chromosome.sh

• You can check the status of your jobs:

For PBS, using qstat:

qstat -u your_username

For SLURM, using squeue:

squeue -u your_username

Job Arrays (PBS vs SLURM)

• Some scripts in this tutorial use job arrays, one job per chromosome:

  • 03_Imputation/04_Convert_and_filter_imputed_genotype_data.sh
  • 04_Ancestry_Checks/02_Tidy_1000G_data.sh
  • 04_Ancestry_Checks/03_Merge_1000G_with_tutorial_data_and_prune.sh

• The scripts are written using PBS job scheduling

PBS job arrays

• The PBS scheduler directive uses #PBS -J to define the range of array jobs, for example:

#!/bin/bash
#PBS -l walltime=01:00:00
#PBS -l mem=80GB
#PBS -l ncpus=1
#PBS -J 1-22

• Within the script, the array index for the current task is accessed using PBS_ARRAY_INDEX, for example:

chr=${PBS_ARRAY_INDEX}

• Another example from this tutorial:

files=(${directory}03_Imputation/Genotype_imputation_results/chr_*.zip)
chr=$(basename "${files[$PBS_ARRAY_INDEX]}" .zip)

SLURM job arrays

• If your HPC system uses SLURM, you must modify these scripts

• Open the script using 'nano' (or a similar editor) and replace the PBS array directive with:

#!/bin/bash
#SBATCH --time=01:00:00
#SBATCH --mem=80GB
#SBATCH --cpus-per-task=1
#SBATCH --array=1-22

• And inside the script replace PBS_ARRAY_INDEX with the SLURM equivalent:

chr=${SLURM_ARRAY_TASK_ID}

files=(${directory}03_Imputation/Genotype_imputation_results/chr_*.zip)
chr=$(basename "${files[$SLURM_ARRAY_TASK_ID]}" .zip)