Supplemental Data
Supplemental Methods
Association and Linkage Analysis:
Likelihood analysis, as implemented in jPAP (14), was used to test T2D, onset age, and BMI for association with each SNP and for genome-wide linkage. For all analyses we excluded affected individuals with onset before age 15 and both affected and unaffected individuals with BMI<14 or BMI>55. Sample size for the association analyses ranged from 1,222 to 1,344 for analysis of T2D and BMI and from 999 to 1,082 for onset age, depending on the number of genotypes available for a given SNP, and equaled the higher number for the linkage analysis.
T2D risk was modeled to account for age at onset in affected pedigree members, while allowing for censored observations, through a modification of the age-of-onset regressive logistic model (34), also known as the age at diagnosis regressive model. Let W represent age at onset or age last examined if unaffected, X=0/1 for male/female, and Y=log (BMI). The logit of the probability of T2D equals
logit[(w, x, y, z)] = + (w-80) + x + (y-log(20))
where (w, x, y, z)=Pr(T=1|W=w, X=x, Y=y) denotes the probability of T2D, p=ln(/(1-)) represents lifetime penetrance for a man with BMI of 20kg/m2, exp( represents the annual OR due to age, exp() represents the female/male OR, exp () represents the OR for a 1-unit increase in log (BMI). Parameter p varies from 0 to 1; parameter varies from >1 to parameters , , and vary from to . Previous implementations of this model assumed the logistic probability density function at onset age for affected individuals and its integral beyond current age for unaffected individuals; the implementation in jPAP substituted, for affected individuals, an integral to one year beyond onset age and applied the inverse normal distribution function to obtain ranges on the liability scale of a polygenic component with heritability h2. The parameters were estimated while correcting from ascertainment through an affected sib pair and, except for heritability, fixed at these estimates in association and variance components linkage analyses.
Onset age and BMI were each modeled as a normal density with mean and standard deviation . Gender was included as a covariate for onset age and gender and age were included as covariates for BMI. Analyses were repeated with BMI as a covariate for onset age and with T2D as a covariate for BMI. Heritability h2 represented the proportion of the variance attributed to polygenes.
Association Analysis
Each of the 5,914 SNPs was tested for association by adding to the models described above a covariate coded as 0/1/2 corresponding to the 3 genotypes, thereby assuming additivity. Then an approximate chi-square statistic was computed as twice the natural logarithm of the ratio of the maximum likelihood of the model with SNP effects included to the maximum likelihood of the submodel without SNP effects; a P-value for one degree of freedom was obtained.
Variance Components Linkage Analysis
Linkage was tested by adding a quantitative trait locus (QTL) as another component of the variance, in addition to the heritability, to the models described above. The lod score at each location was computed as the common logarithm of the maximized likelihood with a QTL effect included divided by the maximized likelihood of the submodel excluding the QTL effect.
The multipoint identity by descent (IBD) probabilities for the variance components linkage analyses were computed by Merlin (13) using the option to identify clusters of closely-linked SNPs with user-defined linkage disequilibrium (LD) between pairs, then assume no recombination within clusters (missing genotypes assigned for obligate recombinants) and no LD between clusters. In the results presented, LD was defined as r2>0.7 (338 SNPs in 147 clusters). Linkage analyses of T2D, onset age, and BMI were repeated for IBD probabilities computed defining clusters for r2>0.5 (462 SNPs in 202 clusters) and r2>0.25 (770 SNPs in 332 clusters); the differences in lod scores were minimal.
Figure S1: Information Content of Microsatellite and SNP marker maps
Information content is shown for Arkansas families tested by microsatellite (red line) or 5914 SNPs (black line).
Table S1: Relationships in GENNID African-American Families
Relationship / NumberFull sibling / 1021
Half siblings / 267
Monozygotic twins / 2
Parent – offspring / 306
Grandparent – grandchild / 14
Avuncular / 43
Half-avuncular / 25
First cousin / 6
Half first-cousin / 3
First cousin once removed / 2
Great avuncular / 4
Half great-avuncular / 2
Table S2: NPL LOD scores for Diabetes Linkage Over 1.0
Chr / Location(cM) / Nearest
Marker / Peak
LOD
1 / 274.00 / rs2027432 / 1.21
3 / 8.96 / rs2290610 / 1.40
6 / 24.00 / rs760892 / 1.20
6 / 90.00 / rs1398576 / 1.00
7 / 27.47 / rs1367781 / 1.73
7 / 77.47 / rs1532083 / 1.19
7 / 151.47 / rs17229 / 1.01
9 / 0.00 / rs1532310 / 1.17
10 / 25.18 / rs827626 / 1.36
11 / 74.00 / rs4255548 / 1.17
13 / 20.00 / rs2491222 / 1.63
14 / 60.59 / rs1253113 / 1.30
16 / 51.14 / rs2078274 / 1.84
16 / 99.14 / rs424074 / 1.08
22 / 40.00 / rs138777 / 1.08
Table 2S shows NPL scores over 1.0 for all families based on type 2 diabetes as the phenotype.
Table S3: Significant Admixture Signals in Order of Decreasing Significance
Chromosome / Location (cM) / Number / P12 / 90.12 / rs1565728 / 0.000316
1 / 192.24 / rs1007460 / 0.00630
11 / 137.93 / rs762827 / 0.0110
15 / 58.39 / rs347117 / 0.0121
9 / 28.49 / rs7866589 / 0.0137
6 / 118.97 / rs1541317 / 0.0186
10 / 99.34 / rs703990 / 0.0194
17 / 62.98 / rs226426 / 0.0217
8 / 30.02 / rs3739406 / 0.0274
19 / 0.00 / rs2012035 / 0.0322
5 / 12.08 / rs816480 / 0.0413
7 / 170.36 / rs1735093 / 0.0417
13 / 71.64 / rs953109 / 0.0422
Table S4: Admixture estimates by study site
Site / N / Minimum / Maximum / AverageAL / 49 / 0.050 / 0.254 / 0.115
AR / 233 / 0.022 / 0.380 / 0.132
CH / 396 / 0.034 / 0.568 / 0.165
CO / 54 / 0.096 / 0.335 / 0.201
LA / 123 / 0.012 / 0.489 / 0.137
LX / 67 / 0.045 / 0.432 / 0.193
MD / 99 / 0.019 / 0.421 / 0.149
NC / 149 / 0.029 / 0.614 / 0.169
SC / 131 / 0.010 / 0.232 / 0.069
SL / 149 / 0.034 / 0.502 / 0.158
Table S4 provides admixture estimates (minimum and maximum observed, average) for each of the 10 centers as described in Table 1 and in the text.
Table S5: SNPs Associated with type 2 diabetes, body mass index, and diabetes age of diagnosis
Chromosome / Location (cM) / Trait / Lod/P-Value / Associated/ Closest SNP / Adjusted for:1 / 14 / T2D / 0.000713 / rs8627
1 / 41 / T2D / 0.000694 / rs861212
1 / 80 / T2D / 0.00189 / rs725330
1 / 211 / T2D / 0.000947 / rs1781014
2 / 56 / AOD / 0.00175 / rs212703 / BMI
2 / 117 / T2D / 0.00197 / rs1051783
2 / 207 / AOD / 0.00105 / rs1396828
2 / 207 / AOD / 0.00161 / rs1396828 / BMI
2 / 241 / BMI / 0.000100 / rs838715
2 / 241 / BMI / 0.000126 / rs838715 / T2D
3 / 37 / T2D / 0.000969 / rs895941
3 / 37 / AOD / 0.00194 / rs895941 / BMI
3 / 70 / T2D / 0.00177 / rs1520483
3 / 182 / BMI / 0.00109 / rs1349838 / T2D
3 / 182 / BMI / 0.00139 / rs1349838
4 / 21 / BMI / 0.000874 / rs6845573 / T2D
4 / 21 / BMI / 0.000944 / rs6845573
5 / 9 / BMI / 0.000858 / rs372169
5 / 9 / BMI / 0.00172 / rs372169 / T2D
5 / 123 / AOD / 0.000168 / rs1459085
5 / 123 / AOD / 0.000230 / rs1459085 / BMI
5 / 123 / T2D / 0.000556 / rs1459085
6 / 51 / AOD / 0.00154 / rs169679 / BMI
6 / 97 / BMI / 0.00230 / rs1055403
6 / 113 / T2D / 0.00143 / rs217532
6 / 114 / T2D / 0.000700 / rs736830
6 / 114 / T2D / 0.00131 / rs1046943
6 / 164 / BMI / 0.00174 / rs231958 / T2D
6 / 164 / BMI / 0.00191 / rs231958
7 / 7 / T2D / 0.000959 / rs1044701
7 / 110 / AOD / 0.00132 / rs1865472 / BMI
7 / 110 / AOD / 0.00191 / rs1865472
7 / 155 / AOD / 0.00102 / rs1024676
7 / 155 / AOD / 0.00187 / rs1024676 / BMI
8 / 59 / AOD / 0.00177 / rs1451687
8 / 82 / T2D / 0.000231 / rs166537
8 / 98 / BMI / 0.000133 / rs1483457 / T2D
8 / 98 / BMI / 0.000152 / rs1483457
9 / 5 / BMI / 0.000413 / rs1331829 / T2D
9 / 5 / BMI / 0.000712 / rs1331829
9 / 5 / T2D / 0.00101 / rs1535842
9 / 88 / AOD / 0.0000510 / rs729958
9 / 88 / AOD / 0.0000796 / rs729958 / BMI
9 / 88 / T2D / 0.000306 / rs729958
9 / 142 / BMI / 0.000332 / rs735107
9 / 142 / BMI / 0.000387 / rs735107 / T2D
11 / 74 / BMI / 0.00176 / rs7122786 / T2D
11 / 74 / BMI / 0.00231 / rs7122786
12 / 24 / BMI / 0.00135 / rs226376
13 / 20 / T2D / 0.000485 / rs2491222
16 / 56 / T2D / 0.00144 / rs11901
16 / 56 / AOD / 0.00196 / rs11901
16 / 58 / AOD / 0.00174 / rs1978320
16 / 58 / AOD / 0.00199 / rs1978320 / BMI
17 / 43 / BMI / 0.000802 / rs2058237
17 / 43 / BMI / 0.00162 / rs2058237 / T2D
18 / 22 / AOD / 0.000211 / rs1941487
18 / 22 / AOD / 0.000278 / rs1941487 / BMI
18 / 115 / T2D / 0.000801 / rs1893828
19 / 15 / T2D / 0.000122 / rs741923
19 / 49 / AOD / 0.00146 / rs1548802 / BMI
19 / 49 / AOD / 0.00198 / rs1548802
19 / 49 / AOD / 0.00165 / rs2009234 / BMI
19 / 49 / AOD / 0.00199 / rs2009234
22 / 44 / AOD / 0.00196 / rs5756477
22 / 47 / BMI / 0.000818 / rs138383
22 / 47 / BMI / 0.000827 / rs138383 / T2D
Table S5 shows association p values from calculated using variance components analysis. In addition to unadjusted body mass index (BMI), type 2 diabetes (T2D), and age of diagnosis (AOD), AOD and BMI were adjusted for BMI and T2D, respectively.
1