Supplementary Figures
Supplementary Figure 1. Long-range LD in the 12q13.13 region.
Data are from HapMap3 as implemented in Haploview. D’ is shown above and r2 below. Standard Haploview colour shading applies.
Supplementary Tables
Supplementary Table 1. Haplotype association analysis in the full sample set at rs6691170 and rs6687758.
rs6691170 allele is shown first and rs6687758 allele second. Estimated haplotype frequencies from PLINK are given. OR=odds ratio, P=P value from haplotype association test. Note that the high-low and low-high risk haplotypes TA and GG have risks intermediate between the other haplotypes. The low-high haplotype is rare.
Risk / HAPLOTYPE / Freq. cases / Freq. Controls / OR / PHigh/high / TG / 0.176 / 0.159 / 1.15 / 1.51x10-13
Low/high / GG / 0.036 / 0.038 / 0.96 / 0.289
High/low / TA / 0.203 / 0.200 / 1.02 / 0.353
Low/low / GA / 0.585 / 0.603 / 0.92 / 3.12x10-9
The Table below shows equivalent data relative to the reference haplotype GA.
Haplotype / No. cases / No. controls / OR / PTG / 7809 / 7751 / 1.14 / 3.9x10-15
GG / 1594 / 1849 / 0.98 / 0.512
TA / 9006 / 9748 / 1.05 / 0.007
GA / 25962 / 29412 / Ref
Supplementary Table 2. Selected individual SNP associations in 1q42 region in UK1, Scotland1 and VQ58 data sets.
Genotype counts in cases and controls are given, together with individual study odds ratio. The risk alleles are T, G, A, G and C for rs6691170, rs6687758, rs11118883, rs12726661 and rs11577023 respectively. Note that the risk allele for rs12726661 is different from that in the logistic regression analysis shown in Table 2 for reasons explained in the main text. Meta-analysis odds ratio (ORmeta), P and I2 measure of heterogeneity are also provided. Note that these analyses are based on the allelic test after direct genotyping or from binning of genotypes from the Impute output. In the latter case, for >98% of samples, predicted genotypes could be easily binned on the basis of clusters; other genotypes were omitted.
Supplementary Table 3. Haplotype association analysis in the full sample set at rs7136702 and rs11169552.
rs7136702 allele is shown first and rs11169552 allele second. Estimated haplotype frequencies from PLINK are given. OR=odds ratio, P=P value from haplotype association test. Note that the low-high risk haplotype CC has risks intermediate between the other haplotypes. The high-low haplotype is rare.
Risk / HAPLOTYPE / Freq. cases / Freq. controls / OR / PHigh/low / TT / 0.021 / 0.022 / 0.90 / 0.069
Low/low / CT / 0.233 / 0.251 / 0.90 / 3.62x10-10
High/high / TC / 0.348 / 0.329 / 1.09 / 8.92x10-9
Low/high / CC / 0.398 / 0.397 / 1.00 / 0.733
The Table below shows equivalent data relative to the reference haplotype CT.
Haplotype / No. cases / No. controls / OR / PTT / 918 / 1070 / 1.03 / 0.578
CT / 10226 / 12234 / Ref
TC / 15281 / 16047 / 1.14 / 2.9x10-38
CC / 17477 / 19361 / 1.08 / 5.9x10-6
Supplementary Table 4. Selected individual SNP associations in 12q13.13 region in UK1, Scotland1 and VQ58 data sets.
Legend is as for Supplementary Table 2, except that the risk alleles are T, C, C, A, G and C for rs7136702, rs11169552, rs3184122, rs35031884, rs7972465 and rs706793 respectively.
Supplementary Table 5. Haplotype analysis in the 12q13.13 region for rs706793, rs7972465 and rs11169552.
The legend is as for Table 4, except that individual haplotypes’ odds ratios and P values are relative to reference haplotype TTT.
Haplotype / No. cases / No. controls / OR / PTTT / 602 / 988 / Ref
CTT / 920 / 1428 / 1.06 / 0.404
CGC / 2502 / 3038 / 1.31 / 2.3x10-7
TTC / 2026 / 2910 / 1.14 / 0.024
CTC / 482 / 648 / 1.22 / 0.012
Supplementary Table 6. Haplotype counts from PLINK analysis in cases and controls at rs706793, rs7972465 and rs11169552.
The upper Table shows test results comparing haplotype counts in cases and controls after fixing the high-risk alleles at rs706793 and rs11169552 and letting rs7972465 vary; the lower Table shows the same after fixing the low-risk allele at rs7972465 and letting rs706793 and rs11169552 vary. Note that the very low frequency of haplotypes containing the G allele at rs7972465 other than CGC makes the other potential similar tests impossible. The significant difference in alleles frequencies when rs706793 and rs11169552 vary is consistent with these SNPs (or SNPs in strong LD with them) affecting risk, whereas there is no significant effect when rs7972465 alleles vary.
Haplotype / CGC / CTCCases / 2509 / 484
Controls / 3151 / 650
P=0.31, Fisher’s exact test
Haplotype / CTC / TTTCases / 484 / 605
Controls / 650 / 993
P=0.012, Fisher’s exact test
Supplementary Table 7. Associations at rs706793 in 7 data sets..
The SNP was genotyped in all series.
Supplementary Table 8. SNPs in LD (r2>0.70) with rs706793, rs7972465 and rs11169552.
The data were derived from the 1000 genomes project via SNAP (http://www.broadinstitute.org/mpg/snap/ldsearch.php) and are restricted to proxy SNPs within 500kb of the test SNP. A search outside these regions revealed no other SNPs with LD as high as r2=0.70, although others were in non-trivial LD (up to r2=0.50). Functional annotation is provided for those SNPs with evidence to support possible effects on protein function, gene expression or other forms of regulation.
7