European Human Genetics Conference 2007 June 16 – 19, 2007 ...
European Human Genetics Conference 2007 June 16 – 19, 2007 ...
European Human Genetics Conference 2007 June 16 – 19, 2007 ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Concurrent Sessions<br />
power of this method is much higher compared to TDT-based methods.<br />
When pedigree structure is at least partly known, utilisation of this<br />
information allows even more powerful analysis. The baseline method<br />
is extended to include covariates, interactions and more complex genetic<br />
effects, such as parent-of-origin and epistasis. We demonstrate<br />
the utility of the newly developed methods using real data from a young<br />
genetically isolated population.<br />
C44. Optimizing information for linkage genomescreen in a large<br />
and inbred pedigree with a high density SNP map<br />
C. Bellenguez 1,2 , C. Ober 3 , C. Bourgain 2,1 ;<br />
1 Univ Paris-Sud, UMR-S535, Villejuif F-94817, France, 2 INSERM, Villejuif F-<br />
94817, France, 3 Department of <strong>Human</strong> <strong>Genetics</strong>, The University of Chicago,<br />
Chicago, IL, United States.<br />
Large genealogies available in isolated populations are potentially<br />
very informative for linkage analyses, in particular when considering<br />
high density SNP maps.<br />
Here, we investigate different methodological aspects of linkage analysis<br />
in a large and inbred 1840-member Hutterite pedigree, phenotyped<br />
for asthma and asthma-related traits. Starting with a 5cM microsatellite<br />
map, we first identify genome-wide significant regions (1p13, 1p31,<br />
5q33, 6q21-23, 12q14, 13q13 and 14p11-q11), after a carefully optimized<br />
breaking of the pedigree. Our approach is interesting since<br />
these regions had not been significantly detected in previous analyses<br />
of the same dataset but have been described by others.<br />
We further analyse these linked regions using a 500K SNP map. Because<br />
SNPs are much closer than microsatellite markers, they present<br />
important linkage disequilibrium (LD), which bias classical nonparametric<br />
multipoint analyses. This problem is even stronger in population<br />
isolates where LD extends over larger regions with a more stochastic<br />
pattern. The only method that models LD in the NPL analysis is limited<br />
in both the pedigree size and the number of markers (Abecasis and<br />
Wigginton, 2005) and therefore could not be used. Instead, we propose<br />
methods that identify sets of SNPs with maximum linkage information<br />
content in our pedigree and no LD-driven bias. Both algorithms<br />
that directly remove pairs of SNPs in high LD and clustering methods<br />
are evaluated. Preliminary results suggest that in such population<br />
isolates, a careful marker selection is necessary to obtain information<br />
content for linkage higher with SNPs than that already available with<br />
microsatellites.<br />
C45. Epistasis for physiological variables in admixed<br />
populations<br />
M. Fenger 1 , A. Linneberg 2 , O. Pedersen 3 , T. Hansen 4 , T. Jøregensen 2 ;<br />
1 University of Copenhagen, Hvidovre, Denmark, 2 Research Centre for Prevention<br />
and Health, Glostrup, Denmark, 3 Steno Diabetes Center, Gentofte, Denmark,<br />
4 Steno Diabetes Centre, Gentofte, Denmark.<br />
The metabolic syndrome and diabetes mellitus are highly influenced<br />
by genetic factors. These conditions are heterogenous and polygenic<br />
in nature for the vast majority of cases, each gene supposed to have<br />
only a minor impact of the phenotypic variance in the general population.<br />
Therefor large populations are required to identify and elucidate the<br />
genetic structure of these conditions. A major obstacle is the heterogeneity<br />
of most study populations, i.e. the populations consist of<br />
a mixture of yet unidentified physiological and genetic homogenous<br />
subpopulations. To define these subpopulations a latent class analysis<br />
is performed. This was done in a structural equation modelling framework.<br />
Basic physiological variables previously shown to be related to<br />
the metabolic syndrome were included in analysis as indicators and<br />
covariates. No genetic model is assumed. This model defined <strong>19</strong> subpopulations<br />
with an entropy measure of approximately 0.9. Less than<br />
0.1%. of all possible single-gene heritabilities were present. In contrast,<br />
including two-gene interactions revealed that 28 of the 30 genetic<br />
markers are involved in one and usually several epistatic interactions.<br />
On average approximately 1/3 of all possible two-gene interactions<br />
were significant for all the traits included in the model. Most notably,<br />
stratifying the basic study population revealed approximately 30%<br />
more interactions masked in the physiological mixed study population.<br />
These results supports the notion that genome wide single-gene association<br />
studies generally will be futile. For a genome wide screening to<br />
be successful the populations should be physiologically stratified and<br />
at least two-gene interactions should be included.<br />
C46. A powerful approach to detect parent-of-origin effects in<br />
whole-genome association scans of quantitative traits<br />
N. M. Belonogova 1,2 , Y. S. Aulchenko 1,3 ;<br />
1 Institute of Cytology & <strong>Genetics</strong> SD RAS, Novosibirsk, Russian Federation,<br />
2 Department of Cytology & <strong>Genetics</strong>, Novosibirsk State University, Novosibirsk,<br />
Russian Federation, 3 Department of Epidemiology & Biostatistics, Erasmus<br />
MC, 3000 CA Rotterdam, The Netherlands.<br />
Parent-of-origin effect (POE) is a widespread genetic phenomenon extensively<br />
studied in model animals. For the genes exhibiting POE, the<br />
trait value in a heterozygous offspring depends on the parental origin<br />
of the alleles. There exist evidences that POE is important for such human<br />
traits as weight, type 2 diabetes, and others. For these traits the<br />
power of genome-wide association (GWA) analysis may be increased<br />
by incorporating POE into the analysis model.<br />
To study POE in GWA scans of human quantitative traits, different variants<br />
of TDT-based methods may be applied. These, however, exploit<br />
only within-family variation in order to avoid population stratification<br />
bias. In homogeneous populations methods utilizing both within- and<br />
between-family variation, such as the measured genotype (MG) approach,<br />
are shown to have greater power. In MG analysis, genetic<br />
polymorphism is included as a fixed effect or covariate in a mixed<br />
linear model along with a polygenic component. To utilize POE information,<br />
we suggest using probabilities indicating parental origin as a<br />
covariate in the MG analysis. We suggest a fast approximation to the<br />
full MG analysis, which is suitable for the analysis of GWA scans and<br />
a method to estimate probabilities of parental origin in arbitrary complex<br />
pedigrees. Using simulated data, we compare our approach to the<br />
TDT-based methods. We show that our approach is more powerful.<br />
For many scenarios the sample size can be reduced two times when<br />
using our approach and the same significance level as by using the<br />
TDT-based methods is still achieved.<br />
C47. Quantifying the increase in human individual genome-wide<br />
heterozygosity through isolate break-up and admixture<br />
I. Rudan 1 , Z. Biloglav 2 , A. Vorko-Jovic 2 , I. Kolcic 2 , O. Polasek 2 , L. Zgaga 2 , T.<br />
Zemunik 3 , R. Mulic 3 , D. Ropac 3 , D. Stojanovic 4 , D. Puntaric 5 , R. Stevanovic 6 , H.<br />
Campbell 1 ;<br />
1 Faculty of Medicine, Edinburgh, United Kingdom, 2 Faculty of Medicine, Zagreb,<br />
Croatia, 3 Faculty of Medicine, Split, Croatia, 4 Faculty of Medicine, Rijeka, Croatia,<br />
5 Faculty of Medicine, Osijek, Croatia, 6 Institute for Public Health, Zagreb,<br />
Croatia.<br />
Aim. The human population is undergoing a major transition from a<br />
metapopulation structure (subdivision in many relatively isolated communities)<br />
to a more admixed structure such as that which occurs in<br />
large cities. We attempted to quantify the magnitude of increase in<br />
average individual genome-wide heterozygosity (IGWH) that has occurred<br />
through isolate break-up and admixture.<br />
Materials and methods. We sampled 100 examinees from 9 isolated<br />
villages on 5 Croatian islands, and an additional 101 immigrants<br />
into those villages. Out of the larger sample of 1001 examinees, we<br />
carefully selected 6 samples each of 23 individuals with a predicted<br />
increasing level of genome-wide heterozygosity, which was then measured<br />
studied using 1,200 STR markers. The first sample was from<br />
the most isolated island, the second from less isolated island, the third<br />
from the least isolated island, the fourth included individuals admixed<br />
between villages, the fifth immigrants from mainland cities and the<br />
sixth extremely outbred individuals.<br />
Results: Relative to the mean IGWH estimate in the first sample, the<br />
increase in the percentage of the genome that is heterozygous in the<br />
remaining 5 samples was found to be: 2.3%, 3.8%, 5.8%, 6.7% and<br />
7.9%. These differences are statistically highly significant (p