Supplementary Figure 1 Dosage correlation between imputed and genotyped alleles Imputed dosages (0 to 2) of 2-digit alleles (red) and 4-digit alleles

Transcription

1 Supplementary Figure 1 Dosage correlation between imputed and genotyped alleles Imputed dosages (0 to 2) of 2-digit alleles (red) and 4-digit alleles (green) of (A) HLA-A, HLA-B, (C) HLA-C, (D) HLA-DQA1, (E) HLA-DQB1, and (F) HLA-DRB1 in 61 East Asian HapMap individuals were compared with the actual dosage (0, 1 or 2). The Pearson's correlation coefficient between the imputed and actual dosages of each allele was plotted according to its allelic frequency. (B)

2 Supplementary Figure 2 Summary of genome-wide SNP datasets

3 Supplementary Figure 3 Principal component analysis for Korean subjects. From a principal component (PC) analysis, we obtained the 10 PCs to adjust population stratification in all statistical analysis. The top 2 PCs were used for plotting the result of PC analysis for (A) Korean data only and (B) Korean and HapMap populations (CEU, CHB+JPT, and YRI).

4 Supplementary Figure 4 Dosage correlation between imputed and genotyped alleles of HLA-DRB1 in study subjects Imputed dosages (0 to 2) of 2-digit alleles (red) and 4-digit alleles (green) of HLA-DRB1 in a subset of the study population (n=1,306) were compared with the actual dosage (0, 1 or 2). The Pearson's correlation coefficient between the imputed and actual dosages of each allele was plotted according to its allelic frequency.

5 Supplementary Figure 5 Correlation of effect estimates of the amino acid haplotype between Korean and Japanese. Japanese data was obtained from Furukawa H et al. (PLoS One 2014;9:e87792; Table 1). All classical alleles was recategorized by the amino acid haplotype Crude ORs were calculated by 2x2 contingency tables according to the presence or absence of haplotypes in cases and controls. The gray line indicates Y=X and green line is a trend line of the data points.

6 Supplementary Figure 6 Correlation of effect estimates of the amino acid haplotype between Korean and European European data was obtained from Morris DL et al (Am J Hum Genet 2012;91: ). All classical alleles was recategorized by the amino acid haplotype Because the paper reported only effect estimate of each classical allele, approximate OR was obtained by performing a meta-analysis for the effect estimates of the classical alleles with a same amino acid haplotype. The gray line indicates Y=X and green line is a trend line of the data points.

7

8 Supplementary Figure 7 Distinct residue effects at amino acid positions 11 and 13 in systemic lupus erythematosus versus other diseases Each odds ratio of residues at the HLA-DRβ1 amino acid positions 11 and 13 in (A-B) European anti-citrullinated protein autoantibody (ACPA) -positive rheumatoid arthritis (RA), (C-D) Asian ACPA-positive RA, (E-F) European ACPA-negative RA, or (G-H) follicular lymphoma (FL) was plotted against to the odds ratio of the matched residue in systemic lupus erythematosus (SLE). Error bars display their 95% confidence intervals and green lines are trend lines of the data points. There were highly discordant trend between the diseases : discordance P = between European ACPA-positive RA and SLE at position 11; discordance P = between European ACPA-positive RA and SLE at position 13; discordance P = between Asian ACPA-positive RA and SLE at position 11; discordance P = between Asian ACPA-positive RA and SLE at position 13; discordance P = between European ACPA-negative RA and SLE at position 11; discordance P = between European ACPA-negative RA and SLE at position 13. *discordance P values between FL and SLE were not calculated due to lack of raw FL data.

9 Supplementary Table 1 Concordance rates between genotyped and imputed HLA alleles of 61 CHB+JPT HapMap individuals Reference panel constructed from Korean* Southeast Asian** X X HLA genes European*** Level A B C DRB1 DQA1 DQB1 Mean 2-digit NA digit NA X X X 2-digit digit digit digit digit digit * Korean HLA reference panel described in Kim K, Bang SY, Lee HS, et al. Construction and Application of a Korean Reference Panel for Imputing Classical Alleles and Amino Acids of Human Leukocyte Antigen Genes. PLoS One 2014; in press. ** Southeast HLA reference panel described in Pillai NE, Okada Y, Saw WY, et al. Predicting HLA alleles from high-resolution SNP data in three Southeast Asian populations. Hum. Mol. Genet ; doi: /hmg/ddu387. *** European HLA reference panel was described in Jia X, Han B, Onengut-Gumuscu S, et al. Imputing amino acid polymorphisms in human leukocyte antigens. PLoS One 2013; doi: /journal.pone