Supplemental Material for Sharma et al., 2018
Figure S1: Example of Illumina 8k Infinium SNP assay showing consistency between manual genotype calling using Illumina Genome Studio software and automated genotype calling using an R package fitTetra.
Figure S2: Plot of the principal components analysis (PCA) performed on the genomic relationship matrix of tetraploid and diploid clones. (a) cluster of cv. Adirondack used as a control in each of the 16 sets of 24 samples genotyped, (b) tetraploid clones, (c) Solanum verrucosum and S. chacoense , (d) S. pinnatisectum and S. bulbocastanum, (e) cv. Mayan Gold (S. tuberosum group Phureja, S. stenotomum and Doubled Monoploid S. tuberosum group Phureja DM1-3 516 R44 clone (DM).
Figure S3: Pairs plot showing a scatterplot of the relationship between traits, the distribution of individual traits and the correlation between traits. ‘.’, ‘*’, ‘**’, ‘***’ signify 0.05, 0.01, 0.001 and 0 p-value thresholds, respectively.
Figure S4: Linkage disequilibrium (LD) measure r2 plotted versus the physical map distance (Mb) between pairs of SNPs in the association panel located on the chromosomal short arm for all chromosomes except chromosome 2 which is telocentric. The trend line of the nonlinear quantile regression of r2 (90th percentile) versus the physical map distance between the SNP markers is depicted in red, dashed blue line depicts the standard LD decay threshold (r2=0.1).
Figure S5: Linkage disequilibrium (LD) measure r2 plotted versus the physical map distance (Mb) between pairs of SNPs in the association panel located on the chromosomal long arm for all 12 chromosomes. The trend line of the nonlinear quantile regression of r2 (90th percentile) versus the physical map distance between the SNP markers is depicted in red, dashed blue line depicts the standard LD decay threshold (r2=0.1).
Figure S6: Screeplot showing how the fidelity of the nMDS (non-metric multidimensional scaling analysis) fit improves as the number of dimensions allowed increases from 2 to 20.
Figure S7: Scatterplot displaying a weak population structure in the potato association panel.
Figure S8: Clustering of tetraploid and diploid genotypes based on Nei (1972) genetic distance. Tetraploid clones are coloured on the basis of their membership to subpopulations identified through nMDS analysis as depicted in Figure 7. Diploid genotypes are shown in black colour.
Figure S9: Screeplot from the Principal Component Analysis (PCA) displaying the number of principal components versus their corresponding eigenvalues.
Figure S10: Bar plots of individual potato genotypes generated by STRUCTURE (Pritchard et al. 2000) using the admixture model assuming allele frequencies are correlated between inferred populations. The analysis was done using 5,718 SNPs. Coloured segments in each column (341 in total) represent membership probabilities of the corresponding clone for the inferred five subpopulations (k=5).
Figure S11: non-metric multidimensional scaling analysis (nMDS) of 341 tetraploid potato clones included in the association panel using 5,157 SNPs. The individual clones are coloured on the basis of their membership to subpopulations identified through nMDS analysis as depicted in Figure 7. The membership details of these clones to their respective subpopulations are presented in Table S2. Each clone is annotated according to its country of origin (AUT, Austria; CAN, Canada; DEU, Germany; DNK, Denmark; ESP, Spain; FRA, France; GBR, United Kingdom; HUN, Hungary; IND, India; IRL, Ireland; ITA, Italy; NLD, Netherlands; NZL, New Zealand; USA, United States of America).
Figure S12: Principal component analysis of 341 tetraploid potato clones included in the association panel using 5,157 SNPs. The individual clones are coloured on the basis of their membership to subpopulations identified through nMDS analysis as depicted in Figure 7. The membership details of these clones to their respective subpopulations are presented in Table S2. Each clone is annotated according to its country of origin (AUT, Austria; CAN, Canada; DEU, Germany; DNK, Denmark; ESP, Spain; FRA, France; GBR, United Kingdom; HUN, Hungary; IND, India; IRL, Ireland; ITA, Italy; NLD, Netherlands; NZL, New Zealand; USA, United States of America).
Figure S13: Q-Q plots comparing the inflation of p-values for the Q models with Q-matrices from different population structure methods for all 20 traits and using the additive marker model. Red circles: nMDSQ Model; Green squares: PCAQ Model; Blue diamonds: STRQ Model. Red line indicates p-values under the expected normal distribution. nMDS, non-metric multidimensional scaling; PCA, principal component analysis; STR, STRUCTURE.
Figure S14: Q-Q plots comparing the inflation of p-values for the QK models with Q-matrices from different population structure methods for all 20 traits and using the additive marker model. Red circles: nMDSQK Model; Green squares: PCAQK Model; Blue diamonds: STRQK Model. Red line indicates p-values under the expected normal distribution. nMDS, non-metric multidimensional scaling; PCA, principal component analysis; STR, STRUCTURE.
Table S1: Details of traits phenotyped in each environment.
Table S2: Details of germplasm included in the study.
Table S3: List of SNPs with multiple mapping positions (Hirsch et al. 2013) in the potato genome.
Table S4: Comparison of genotype calls (diploid germplasm only) processed using GenomeStudio (Illumina Software) and fitTetra (R package).
Table S5: Subset of 5,718 SolCAP SNPs present in (a) chromosomal short arm, (b) chromosomal long arm, (c) euchromatin, and (d) heterochromatin.
Table S6: Approximate size of pericentromeric heterochromatin in all 12 chromosomes deduced from Sharma et al. (2013).
Table S7: Pairwise measures of population differentiation (Fst) among five tetraploid subpopulations and one diploid wild germplasm group.
Table S8: Significant marker-trait associations (MTAs) obtained using four main GWAS models.
Table S9: Annotation of 5,718 SolCAP SNPs.
Table S10: Genotype and phenotype data.