1. The effect of using sequence data instead of a lower density SNP chip on a GWAS
EAAP 2017; Tallinn, Estonia
Sanne van den Berg, Roel Veerkamp, Fred van Eeuwijk, Aniek Bouwman, Marcos Lopes, Jérémie Vandenplas

2. Background Phenotypes Understanding genetic architecture
Prior for genomic prediction
Associated
SNPs
GWAS
Genotypes
Sequence data is expected to improve identification of associated SNPs

3. Objective
To investigate the effect of using sequence data instead of a lower density SNP chip on a GWAS
Do we find more QTL using sequence data?
Do we explain more of the genetic variation within lines?
Do we explain more of the genetic variation across lines?

4. Data 12184 Large White pigs 80K: 35,465 SNP 660K: 504,170 SNP
4943 Dutch landrace pigs
80K: 39,109 SNP
660K: selected from imputed sequence data (311,888 SNP)
Trait: Number of teats
Heritability 0.40

5. Imputed sequence data Dutch Landrace: Large White:
26.1M SNPs were imputed with Beagle using a multi-line reference population
Quality control:
remove SNP with accuracy < 0.6
MAF < 0.01
Large White:
660K  sequence
10.2M SNPs
Accuracy = 0.93
Dutch Landrace:
80K  sequence
7.8M SNPs
Accuracy = 0.84

6. GWAS
Software: GCTA
Mixed linear association model using a leave on chromosome out approach
Significance threshold: -log10(P-value) > 5
QTL region: <0.5 MB> from most significant SNP
Yang et al. (2014)
Number of independent chromosome fragments Goddard et al. (2011)
80K
660K
Sequence
Large White
780
650

7. Evaluation of QTL regions
Genomic heritability:
= the variance explained by the genomic relationships based on the most significant SNP in a QTL region
Within line: 3-fold cross validation
Across line: GWAS in Large White  evaluation QTL in Dutch Landrace

8. Manhattan plot: Dutch Landrace
5 QTL with a large effect found by Duijvesteijn et al 2014
291 SNPs
73 QTL on 10 chromosomes
2988 SNPs
129 QTL on 13 chromosomes
54317 SNPs QTL on 16 chromosomes

9. Manhattan plot: Large White
5 QTL with a large effect found by Duijvesteijn et al 2014
1673 SNPs
443 QTL on 10 chromosomes
23106 SNPs
794 QTL on 13 chromosomes
SNPs QTL on 16 chromosomes

10. Evaluation QTL regions Within line
Average genomic heritability
The average genomic heritability increases when increasing from lower density to WGS

11. Evaluation QTL regions Large White  Dutch Landrace
The genomic heritability does not increase when increasing from lower density to WGS
Differences in genetic architecture
Different causal mutations
Differences MAF
Genomic heritability

12. Take home message
To investigate the effect of using sequence data instead of a lower density SNP chip on a GWAS
Using sequence data instead of lower density SNP chip:
Increases the number of QTL
Improves the genomic heritability within line
Does not improve the genomic heritability across line

13. Take home message
To investigate the effect of using sequence data instead of a lower density SNP chip on a GWAS
Using sequence data instead of lower density SNP chip:
Increases the number of QTL
Improves the genomic heritability within line
Does not improve the genomic heritability across line
Thank you for your attention!
Sanne van den Berg