Genome-wide association studies in Canada's Apple Biodiversity Collection

Sophie Watts¹, ‪Zoë Migicovsky¹, Sean Myles*¹

Affiliations
¹Department of Plant, Food, and Environmental Sciences, Faculty of Agriculture, Dalhousie University

---

Table of contents

1. About
   
2. Abstract
   
3. Analysis
   3a. Getting the code
   3b. File architecture
   3c. Description of files
   

1. About

This repository contains data and scripts used to repoduce analyses in the manuscript "Genome-wide association studies in Canada's apple biodiversity collection".

2. Abstract

Apple fruit quality traits such as fruit texture, sugar content, and firmness retention during storage are key targets for breeders. Understanding the genetic control of fruit quality traits can enable the development of genetic markers, useful for marker-assisted breeding of new apple cultivars. We genotyped over 260,000 single nucleotide polymorphisms (SNPs) across 1,054 apple accessions from Canada’s Apple Biodiversity Collection and performed genome-wide association for 21 fruit quality and phenology traits. We identified a locus on chromosome 15 associated with phenolic content and a locus on chromosome 10 that is associated with softening. We demonstrate that the top SNP on chromosome 10 is a better predictor of softening than markers commonly used for marker-assisted breeding of this trait. In addition, we identified a single locus on chromosome 3 that is associated with numerous traits including ripening time, firmness at harvest, and firmness after storage. The top SNP at the chromosome 3 locus is a nonsynonymous mutation within the NAC18.1 transcription factor. Given the association between variation at NAC18.1 and several key traits, we propose a model for the allelic effects at NAC18.1 on apple ripening and softening.

3. Analysis

3a. Getting the code

You can download a copy of all the files in this repository by cloning the git repository:

$ git clone https://github.com/MylesLab/abc-gwas.git

3b. File architecture

├── source
├── data
├── outputs
├── shell scripts
├── GWAS results
└── figures

• The data directory contains all the raw data that was used for this project.
• The outputs directory contains the files generated from the raw data through data curation.
• The source directory contains the scripts used for data curation and performing various analyses of this project.
• The shell scripts directory contains the shell scripts to run code over the command line.
• The figures directory contains the intermediary figures generated for this project. Final figures were assembled using Adobe Illustrator.

3c. Description of files

Source

1. phenotype_curation.Rmd Code for curating phenotype table and outputting a list of phenotypes.
2. pop_analyses Code for visualizing prinicipal components analysis.
3. gwas_script.Rmd script to create shell scripts to filter phenotype and genotype data, create kinship matrices, run PCA.
4. simple_m.R script to run Simple M to calculate the effective number of markers.
5. mlmm_gwas_batch_script_final.R code to run the MLMM GWAS.
6. correlations.R code to run and visualize correlations between phenotypes of interest.
7. top_snp_genotypes.Rmd code to extract the genotypes of the top SNP hits from the GWAS of interest.
8. ripening_model.Rmd Script to create the ripening model figure.
9. boxplots_manhattans.Rmd Code for plotting manhattan plots and boxplots.
10. zoom_plots Script to plot zoom ins of manhattan plots with gene annotations.
11. snp_variance Script to calculate the proportion of variance explained by the top GWAS snps.

Data

1. pheno_meta_data.csv Phenotype data and meta data for all accessions in the ABC from Watts et al. 2021.
2. abc_combined_maf001_sort_vineland_imputed_pheno_hetero90_maf001.nosex File with the apple IDS that have genetic data.
3. abc_combined_maf001_sort_vineland_imputed_pheno_hetero90_maf001.frq Minor allele frequency file for ABC snps.
4. abc_combined_maf001_sort_vineland_imputed_pheno_hetero90_maf001_noctgs_pruned1.txt PC values from PCA, contig snps removed and LD pruned.
5. abc_combined_maf001_sort_vineland_imputed_pheno_hetero90_maf001_noctgs_pruned2.txt Eigen values from PCA, contig snps removed and LD pruned.
6. abc_combined_maf001_sort_vineland_imputed_pheno_hetero90_maf001_het.het Plink file with heterozygosity per individual.
7. top_snps.txt file with the names of the top SNP hits from the GWAS of interest.
8. abc_combined_maf001_sort_vineland_imputed_pheno_hetero90_maf001_top_gwas_snps.raw Genotype file that has been subset to only include the top SNPs of interest from the GWAS.
9. abc_combined_maf001_sort_vineland_imputed_pheno_hetero90_maf001_top_gwas_snps.ped PED genotype file that has been subset to only include the top SNPs of interest from the GWAS.
10. abc_combined_maf001_sort_vineland_imputed_pheno_hetero90_maf001_top_gwas_snps.map MAP genotype file that has been subset to only include the top SNPs of interest from the GWAS.
11. gene_models_20170612.gff3.gz Gene model annotations from the GDDH genome version 1.
12. vineland_snps.txt The list of SNPs that were genotyped using HRM.

Outputs

1. geno_pheno_meta_data.csv Filtered phenotype table that includes 1054 apple IDs that have both phenotype data and genotype data.
2. pheno_list.txt List with names of phenotypes from Watts et al. 2021.
3. pheno_list_2017 List with names of phenotypes used for GWAS.
4. contig_snps_remove.txt List of unanchored SNPs to be removed from SNP table.
5. pc1_vs_phenos.csvThe R-square and p-values from the correlation of phenotypes with PC1.
6. pc2_vs_phenos.csv The R-square and p-values from the correlation of phenotypes with PC2.
7. phenotype_sample_sizes.txt Table with samples sizes for each phenotype.
8. phenotypes4gwas Folder that contains two files for each phenotype: one with a list of apple IDs for that phenotype that have trait data and the second a file with the trait measurements per apple ID for that phenotype.
9. simple_m.out Output from Simple M package that calculates the effective number of markers.
10. top_snp_genos.csv File with the genotypes of the top SNP hits from the GWAS.
11. gene_annotations Folder containing the files with gene annotations surrounding the top GWAS hits.
12. snp_variation.csv R-square values from LMs with top GWAS SNPs and PCs.
13. top_snps_pheno_pcs.csv File with genotypes of the top SNPs from GWAS, pheno data and PCs.
14. ripening_model_summary.csv Median values for traits measurements across the genotypic classes at NAC18.1

Shell Scripts

1. abc_pca.sh script to run PCA for the whole SNP set with TASSEL.
2. genotype_filtering_plink.sh script that contains PLINK commands to filter the ABC MAP and PED to only containing apple IDs for a particular phenotype, applies a MAF filter of 0.01, and outputs a MAP and PED for each phenotype.
3. kinship.sh script that contains the tassel commands to make a kinship matrix for each phenotype.
4. pca.sh script commands to run PCA with tassel for each individual phenotype file.
5. geno_raw.sh script with commands to recode PED and MAP files into .raw files for the MLMM gwas.
6. simple_m.sh script to run simple_m.R
7. run_mlmm_gwas_batch_script_final.sh script to run the MLMM GWAS code.

GWAS results

1. mlmm_pvals Folder containing files for each phenotype with the SNP p-values from the MLMM GWAS.
2. mlmm_qq Folder containing qq-plots from each MLMM GWAS.
3. mlmm_manhattans Folder containing manhattan plots from each MLMM GWAS.
4. rss Folder containing files for each phenotype with the variance explained by the co-factor SNPs at each step of the MLMM GWAS.
5. standard_pvals Folder containing files for each phenotype with the SNP p-values from the standard (MLM) GWAS.
6. standard_qq Folder containing qq-plots from each standard (MLM) GWAS.
7. standard_manhattans Folder containing manhattan plots from each standard (MLM) GWAS.
8. sbatch_command_mlmm.txt commands to excute run_mlmm_gwas_batch_script_final.sh.