Electronic supplementary material
Statistical power
The study had 80% power to detect a susceptibility locus with a locus-specific value of l for siblings with a LOD=1.55 for suggestive evidence of linkage (pointwise p<0.00074) and a LOD=1.75 for significant evidence of linkage (pointwise p<0.000022).
Simulation studies for obtaining empirical genome-wide p value
Critical values were obtained empirically by simulating genotype data when no linked loci were available. To do this, we generated 1,000 replicates using the simulate option in MERLIN under the experimental and data conditions prevailing in our study and analysed those replicates with linear models implemented in MERLIN. We then counted the number of times a certain threshold was crossed by the resulting linkage trace. In our case we determined the locus number expected by chance for the results on chromosomes 1 and 16 (LOD=1.49/1.85). In addition, we used the locus counting approach, as proposed by Wiltshire and co-workers [1], to determine empirical LOD scores at which, in a genome scan, statistical evidence would be expected to occur once, 0.05 times and 0.001 times at random, corresponding to suggestive, significant (genome-wide p=0.05) and highly significant linkage (genome-wide p=0.001) according to the guidelines of Lander and Kruglyak [2] (ESM Table 4). For our procedure it was crucial to count only independent (unlinked) regions. Regions were considered as independent, if their respective maxima were separated by ≥55 cM (Haldane map function) as proposed by Ott in 1999 [3].
References
1. Wiltshire S, Cardon LR, McCarthy MI (2002) Evaluating the results of genomewide linkage scans of complex traits by locus counting. Am J Hum Genet 71: 1175–1182
2. Lander E, Kruglyak L (1995) Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet 11: 241–247
3. Ott J (1999) Analysis of human genetic linkage. The Johns Hopkins University Press, Baltimore.
4. Cox NJ, Frigge M, Nicolae DL et al (1999) Loci on chromosomes 2 (NIDDM1) and 15 interact to increase susceptibility to diabetes in Mexican Americans. Nat Genet 21: 213–215
ESM Table 1 Results of nonparametric two-point linkage analyses (SPLINK version 1.09).
Marker / CHR / Position (cM) / HET / LOD / Nominal p values / IBD(0) / IBD(1) / IBD(2)D1S3669 / 1 / 37.05 / 0.7874 / 1.08 / 0.0191 / 0.1793 / 0.5000 / 0.3207
D2S1328 / 2 / 132.58 / 0.7634 / 2.02 / 0.0020 / 0.2120 / 0.4241 / 0.3639
D3S3050 / 3 / 14.46 / 0.8359 / 2.87 / 0.0003 / 0.1848 / 0.4442 / 0.3710
D7S1802 / 7 / 33.09 / 0.7389 / 1.17 / 0.0153 / 0.1904 / 0.4784 / 0.3311
D11S1314 / 11 / 73.64 / 0.8102 / 1.04 / 0.0215 / 0.2240 / 0.4480 / 0.3280
D16S403 / 16 / 43.89 / 0.8317 / 1.33 / 0.0104 / 0.1783 / 0.5000 / 0.3217
Only LODs ≥1 are shown. CHR, chromosome; HET, heterozygosity; IBD(0), proportion of affected sib pairs sharing 0 alleles; IBD(1), proportion of affected sib pairs sharing 1 allele; IBD(2), proportion of affected sib pairs sharing two alleles.
ESM Table 2 Results of multipoint linkage analyses using Merlin
D1S1597 / 1 / 29.93 / 1.51 / 0.78 / 0.030
D1S3669 / 1 / 37.05 / 2.08 / 1.49 / 0.004
D1S552 / 1 / 45.33 / 1.85 / 1.15 / 0.011
D1S1622 / 1 / 56.74 / 1.45 / 0.71 / 0.040
D3S3630 / 3 / 10.70 / 1.52 / 0.63 / 0.040
D4S403 / 4 / 25.90 / 1.40 / 0.61 / 0.050
D11S2002 / 11 / 85.48 / 1.51 / 0.76 / 0.030
D16S403 / 16 / 43.89 / 2.47 / 1.85 / 0.002
D16S769 / 16 / 50.60 / 1.99 / 1.24 / 0.008
D21S1260 / 21 / 46.71 / 1.59 / 0.91 / 0.020
Only non-parametric LOD scores >0.59 (nominal p<0.05) are shown. CHR chromosome.
ESM Table 3 Results from analyses of 1,000 replicates using our data structure, analysed with a linear model using MERLIN
0.59 / 8269 / 8.269 / Nominal p=0.05
1.18 / 2193 / 2.193 / Nominal p=0.01
1.49a / 1047 / 1.047 / D1S3669 (37.05 cM)
1.51 / 1005 / 1.005 / Suggestive (our simulations)
1.85 a / 454 / 0.454 / D16S403 (43.89 cM)
2.20 / 217 / 0.217 / Nominal p=0.0007
Suggestive (Lander & Kruglyak [2])
2.74 / 50 / 0.050 / Significant (our simulations)
3.00 / 24 / 0.024 / Nominal p=0.0001
3.63 / 8 / 0.008 / Nominal p=0.00005
Significant (Lander & Kruglyak [2])
4.50 / 1 / 0.001 / Highly significant (our simulations)
5.30 / 0 / 0.000 / Nominal p=0.0000003
Highly significant (Lander & Kruglyak [2])
According to our simulation studies, suggestive, significant or highly significant linkage can be assumed at LOD≥1.51, LOD≥2.74 or LOD 4.50 respectively.
a LOD scores for loci on chromosomes 1 and 16.
IRL regions considered to be independent
ESM Table 4 Results of conditional multipoint linkage analyses using ALLEGRO
Marker / CHR / Position(cM) / Baseline
LOD / Weighted
LOD / χ2 / p value / Conditioned
on STRP / Weight
D4S403–D4S2639 / 4 / 29.88 / 0.63 / 1.20 / 2.632 / NS / D1S3669 / prop
D11S1392 / 11 / 43.16 / 0.03 / 1.28 / 5.763 / NS / D1S3669 / prop
D16S403–D16S769 / 16 / 45.23 / 1.85 / 1.96 / 0.475 / NS / D1S3669 / 0–1
45.23 / 1.85 / 1.45 / 1.843 / NS / D1S3669 / prop
D21S1260 / 21 / 46.71 / 0.91 / 1.35 / 2.025 / NS / D1S3669 / 1–0
D17S2180 / 17 / 66.85 / 0.08 / 1.63 / 7.137 / 0.0075 / D16S403 / prop
D1S552–D1S1622 / 1 / 47.61 / 1.14 / 2.81 / 7.706 / 0.0055 / D17S2180 / 0–1
47.61 / 1.14 / 2.36 / 5.598 / NS / D17S2180 / prop
D1S1679–D1S1589 / 1 / 185.48 / 0.03 / 1.24 / 5.540 / NS / D17S2180 / 0–1
210.39 / 0.06 / 1.66 / 7.370 / 0.0066 / D17S2180 / prop
D4S2366–D4S2639 / 4 / 33.42 / 0.57 / 1.29 / 3.344 / NS / D17S2180 / 0–1
15.85 / 0.50 / 1.40 / 4.160 / NS / D17S2180 / prop
D16S403 / 16 / 43.89 / 1.85 / 2.51 / 3.047 / NS / D17S2180 / 0–1
43.89 / 1.85 / 3.33 / 6.802 / 0.0091 / D17S2180 / prop
Initially, conditioning was performed on the two loci (on chromosomes 1 and 16) with the largest signals in primary analysis. The chromosome 16 conditioning increased evidence for linkage at a locus on chromosome 17, leading us to perform conditioning on chromosome 17. Here, only loci where the weighted LOD was ≥1.18 (nominal p≤0.01) and larger than the baseline LOD are shown. The significance associated with the increased LOD score (ΔLOD) can be determined by using a conservative χ2 test with 1df (χ2=2ln[10][weighted LOD−baseline LOD]) [4]. To take multiple testing into account, including regions that did not show a significant baseline LOD, p values must be lower than 0.01 (shown here), instead of the conventional 0.05. CHR chromosome; STRP short tandem repeat polymorphisms; prop proportional to the evidence for linkage
ESM Fig. 1 a-f Results of conditional nonparametric linkage analyses as introduced by Cox et al. [4] and performed with ALLEGRO. a Multipoint allele-sharing analysis of chromosome 16 weighted by the evidence for linkage at D1S3669 on chromosome 1. No interaction between either chromosomes was observed. b Multipoint allele-sharing analysis of chromosome 1, weighted by the evidence for linkage at D16S403 on chromosome 16 and confirming the result from the previous conditioning. Continuous line (a, b): weight 0–1; dotted line (a, b): weight 1–0; dashed line (a, b): proportional to the evidence for linkage (prop); bold line (a, b): unweighted. c The conditioning on D16S403 revealed another locus on chromosome 17 (D17S2180). Continuous line: D16S403 weight 0–1; dotted line: D16S403 weight 1–0; dashed line D16S403 prop; bold line: D16S403 unweighted. d, e The identification of D17S2180 led to another conditioning on D17S2180. Here we show the results of the multipoint allele-sharing analysis of chromosomes 16 (d) and 1 (e), weighted by the evidence for linkage at D17S2180. Continuous line (d, e): D17S2180 weight 0–1; dotted line (d, e): D17S2180 weight 1–0; dashed line (d, e): D17S2180 prop; bold line (d, e): unweighted. f To test both directions, we performed a multipoint allele-sharing analysis of chromosome 17 weighted by the evidence for linkage at D1S3669. Continuous line: D1S3669 weight 0–1; dotted line: D1S3669 weight 1–0; dashed line: D1S3669 prop; bold line: D1S3669: unweighted.
ESM Fig. 2 Genome-wide LOD plots for conditional linkage analysis. The conditioning was done for markers D1S3669 (a), D16S403 (b) and D17S2180 (c) by: (1) assigning weight 0 to families with 0 or negative linkage scores and weight 1 to families with positive linkage scores (weight 0–1); (2) modelling heterogeneity by assigning weight 1 to families with negative linkage scores and weight 0 to families with 0 or positive linkage scores (weight 1–0); and (3) assigning family-specific weights proportional to the evidence for linkage (weight prop). Grey arrows: regions containing the conditioned markers. Continuous lines: weight 0–1; dotted lines: weight 1–0; dashed lines: prop; bold lines: unweighted
ESM Fig. 1
a
b
c
d
e
f
ESM Fig. 2
a
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 X
Chromosomes
b
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 X
Chromosomes
c
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 X
Chromosomes