Use of a random coefficient regression (RCR) model to estimate growth parameters

We used a random coefficient regression (RCR) model to estimate growth parameters for the time series of observed serum glucose levels in the Replicate 1 of the Genetic Analysis Workshop 13 simulated data. For comparison, a two time-point interval was also selected and the slope between these two observations was calculated. This process yielded four phenotypes: the RCR growth phenotype, a two time-point slope phenotype, and Time 1 and Time 2 serum glucose level phenotypes. These four phenotypes were used for linkage analyses on simulated chromosomes 5, 7, 9, and 21, those chromosomes that contained loci affecting the growth course for serum glucose levels. The linkage analysis of the RCR-derived phenotype showed overwhelming evidence for linkage at one locus (LOD 65.78 on chromosome 5), while showing elevated but nonsignificant LOD scores for two other loci (LOD 1.25 on chromosome 7, LOD 1.10 on chromosome 9), and no evidence of linkage for the final locus. The two time-point slope phenotype showed evidence for linkage at one locus (LOD 4.16 on chromosome 5) but no evidence for linkage at any of the other loci. A parallel cross-sectional approach, using as input phenotypes the endpoints of the two-point slope phenotype, gave strong linkage results for the major locus on chromosome 5 (maximal LOD scores of 17.90 and 27.24 for Time 1 and Time 2, respectively) while showing elevated but nonsignificant linkage results on chromosome 7 (maximal LOD scores of 1.71 and 1.48) and no evidence for linkage at the two remaining loci. The RCR growth parameter showed more power to detect linkage to the major locus than either the cross-sectional or two-point slope approach, but the cross-sectional approach gave a higher maximal LOD score for one of the minor loci.     

Linkage analysis of the genetic determinants of high density lipoprotein concentrations and composition: evidence for involvement of the apolipoprotein A-II and cholesteryl ester transfer protein loci

We have tested for evidence of linkage between the genetic loci determining concentrations and composition of plasma high density lipoproteins (HDL) with the genes for the major apolipoproteins and enzymes participating in lipoprotein metabolism. These genes include those encoding various apolipoproteins (apo), including apoA-I, apoA-II, apoA-IV, apoB, apoC-I, apoC-II, apoC-III, apoE, and apo(a), cholesteryl ester transfer protein (CETP), HDL-binding protein, lipoprotein lipase, and the low density lipoprotein (LDL) receptor. Polymorphisms of these genes, and nearby highly polymorphic simple sequence repeat markers, were examined by quantitative sib-pair linkage analysis in 30 coronary artery disease families consisting of a total of 366 individuals. Evidence for linkage was observed between a marker locus D16S313 linked to the CETP locus and a locus determining plasma HDL-cholesterol concentration (P = 0.002), and the genetic locus for apoA-II and a locus determining the levels of the major apolipoproteins of HDL, apoA-I and apoA-II (P = 0.009 and 0.02, respectively). HDL level was also influenced by the variation at the apo(a) locus on chromosome 6 (P = 0.02). Thus, these data indicate the simultaneous involvement of at least two different genetic loci in the determination of the levels of HDL and its associated lipoproteins.     

Multilocus linkage analysis of affected sib pairs

OBJECTIVE: The conventional affected sib pair methods evaluate the linkage information at a locus by considering only marginal information. We describe a multilocus linkage method that uses both the marginal information and information derived from the possible interactions among several disease loci, thereby increasing the significance of loci with modest effects. METHODS: Our method is based on a statistic that quantifies the linkage information contained in a set of markers. By a marker selection-reduction process, we screen a set of polymorphisms and select a few that seem linked to disease. RESULTS: We test our approach on genome scan data for inflammatory bowel disease (InfBD) and on simulated data. On real data we detect 6 of the 8 known InfBD loci; on simulated data we obtain improvements in power of up to 40% compared to a conventional single-locus method. CONCLUSION: Our extensive simulations and the results on real data show that our method is in general more powerful than single-locus methods in detecting disease loci responsible for complex traits. A further advantage of our approach is that it can be extended to make use of both the linkage and the linkage disequilibrium between disease loci and nearby markers.     

Utility and efficiency of linked marker genes for genetic counseling. II. Identification of linkage phase by offspring phenotypes

For a linked marker locus to be useful for genetic counseling, the counselee must be heterozygous for both disease and marker loci and his or her linkage phase must be known. It is shown that when the phenotypes of the counselee's previous children for the disease and marker loci are known, the linkage phase can often be inferred with a high probability, and thus it is possible to conduct genetic counseling. To evaluate the utility of linked marker genes for genetic counseling, the accuracy of prediction of the risk for a prospective child with a given marker gene to develop the genetic disease and the proportion of families in which a particular marker locus can be used for genetic counseling are studied for X-linked recessive, autosomal dominant, and autosomal recessive diseases. In the case of X-linked genetic diseases, information from children is very useful for determining the linkage phase of the counselee and predicting the genetic disease. In the case of autosomal dominant diseases, not all children are informative, but if the number of children is large, the phenotypes of children are often more informative than the information from grandparents. In the case of autosomal recessive diseases, information from grandparents is usually useless, since they show a normal phenotype for the disease locus. If we use information on the phenotypes of children, however, the linkage phase of the counselee and the risk of a prospective child can be inferred with a high probability. The proportion of informative families depends on the dominance relationship and frequencies of marker alleles, and the number of children. In general, codominant markers are more useful than are dominant markers, and a locus with high heterozygosity is more useful than is a locus with low heterozygosity.     

Unrecognized sequence homologies may confound genome-wide association studies

Genome-wide association studies (GWAS) have become a preferred method to identify new genetic susceptibility loci. This technique aims to understanding the molecular etiology of common diseases, but in many cases, it has led to the identification of loci with no obvious biological relevance. Herein, we show that previously unrecognized sequence homologies have caused single-nucleotide polymorphism (SNP) microarrays to incorrectly associate a phenotype to a given locus when in fact the linkage is to another distant locus. Using genetic differences between male and female subjects as a model to study the effect of one specific genomic region on the whole SNP microarray, we provide strong evidence that the use of standard methods for GWAS can be misleading. We suggest a new systematic quality control step in the biological interpretation of previous and future GWAS.     

Surrogate phenotype definition for alcohol use disorders: a genome-wide search for linkage and association

For the identification of susceptibility loci in complex diseases the choice of the target phenotype is very important. We compared results of genome-wide searches for linkage or for association related to three phenotypes for alcohol use disorder. These are a behavioral score BQ, based on a 12-item questionnaire about drinking behavior and the subject's report of drinking-related health problems, and ERP pattern and ERP magnitude, both derived from the eyes closed resting ERP measures to quantify brain activity. Overall, we were able to identify 11 candidate regions for linkage. Only two regions were found to be related to both BQ and one of the ERP phenotypes. The genome-wide search for association using single-nucleotide polymorphisms did not yield interesting leads.     

A genome-wide linkage scan identifies multiple chromosomal regions influencing serum lipid levels in the population on the Samoan islands

Abnormal lipid levels are important risk factors for cardiovascular diseases. We conducted genome-wide variance component linkage analyses to search for loci influencing total cholesterol (TC), LDL, HDL and triglyceride in families residing in American Samoa and Samoa as well as in a combined sample from the two polities. We adjusted the traits for a number of environmental covariates, such as smoking, alcohol consumption, physical activity, and material lifestyle. We found suggestive univariate linkage with log of the odds (LOD) scores > 3 for LDL on 6p21-p12 (LOD 3.13) in Samoa and on 12q21-q23 (LOD 3.07) in American Samoa. Furthermore, in American Samoa on 12q21, we detected genome-wide linkage (LOD(eq) 3.38) to the bivariate trait TC-LDL. Telomeric of this region, on 12q24, we found suggestive bivariate linkage to TC-HDL (LOD(eq) 3.22) in the combined study sample. In addition, we detected suggestive univariate linkage (LOD 1.9-2.93) on chromosomes 4p-q, 6p, 7q, 9q, 11q, 12q 13q, 15q, 16p, 18q, 19p, 19q and Xq23 and suggestive bivariate linkage (LOD(eq) 2.05-2.62) on chromosomes 6p, 7q, 12p, 12q, and 19p-q. In conclusion, chromosome 6p and 12q may host promising susceptibility loci influencing lipid levels; however, the low degree of overlap between the three study samples strongly encourages further studies of the lipid-related traits.     

Localization and identification of human quantitative trait loci: king harvest has surely come

The scientific process of localization and subsequent identification of genes influencing risk of common diseases is still in its infancy. Initial localization of disease-related loci has traditionally been performed using family-based linkage methods to scan the genome. Early pronouncements of the failure of this approach for common diseases were premature and based on comparing suboptimal linkage designs with overly optimistic and empirically unproven association-based designs. On the contrary, substantial recent progress in the positional cloning of genes influencing such complex phenotypes suggests that modern approaches based around a family-based linkage paradigm will be successful. In particular, the rapidly growing emphasis on the analysis of the genetic basis of quantitative correlates of disease risk represents a novel and promising approach in which initial localization is performed using linkage and subsequent identification utilizes association approaches in positional candidate genes.     

Joint Multipoint Linkage Analysis of Multivariate Qualitative and Quantitative Traits. II. Alcoholism and Event-Related Potentials

The availability of robust quantitative biological markers that are correlated with qualitative psychiatric phenotypes can potentially improve the power of linkage methods to detect quantitative-trait loci influencing psychiatric disorders. We apply a variance-component method for joint multipoint linkage analysis of multivariate discrete and continuous traits to the extended pedigree data from the Collaborative Study on the Genetics of Alcoholism, in a bivariate analysis of qualitative alcoholism phenotypes and quantitative event-related potentials. Joint consideration of the DSM-IV diagnosis of alcoholism and the amplitude of the P300 component of the Cz event-related potential significantly increases the evidence for linkage of these traits to a chromosome 4 region near the class I alcohol dehydrogenase locus ADH3. A likelihood-ratio test for complete pleiotropy is significant, suggesting that the same quantitative-trait locus influences both risk of alcoholism and the amplitude of the P300 component.     

Stratification by CARD15 variant genotype in a genome-wide search for inflammatory bowel disease susceptibility loci

Previously we have conducted a genome-wide search for inflammatory bowel disease susceptibility loci in a large European cohort. Results from this study demonstrated suggestive evidence of linkage to loci at chromosomes 1q, 6p, and 10p and replicated linkages on chromosomes 12 and 16. Recently, NOD2/CARD15 on chromosome 16q12 has been found to be strongly associated with Crohn's disease. In order to determine if there are other loci in the genome that interact with the three associated functional variants in CARD15 (R702W, G908R, 1007fs), we have stratified our large inflammatory bowel disease genome scan cohort by dividing pedigrees into two groups stratified by CARD15 variant genotype. The two pedigree groups were analysed using non-parametric allele sharing methods. The group of pedigrees that contained one of the three CARD15 variants had two suggestive linkage results occurring in 6p (lod = 3.06 at D6S197, IBD phenotype) and 10p (lod=2.29 at D10S197, CD phenotype). In addition, at 16q12 where CARD15 is located, the original genome scan had a peak lod score of 2.18 at D16S415 (CD phenotype). The stratified pedigree cohort containing one of three CARD15 variants had a peak lod score of 0.90 at D16S415 (CD phenotype), accounting for approximately less than half of the genetic evidence for linkage at this locus. This result is in agreement with the existence of a substantial number of private variants at the NOD2/CARD15 locus. Interaction with NOD2/CARD15 needs to be considered in future gene identification efforts on chromosomes 6 and 10.     

Experimental designs for robust detection of effects in genome-wide case-control studies

In genome-wide association studies hundreds of thousands of loci are scanned in thousands of cases and controls, with the goal of identifying genomic loci underpinning disease. This is a challenging statistical problem requiring strong evidence. Only a small proportion of the heritability of common diseases has so far been explained. This "dark matter of the genome" is a subject of much discussion. It is critical to have experimental design criteria that ensure that associations between genomic loci and phenotypes are robustly detected. To ensure associations are robustly detected we require good power (e.g., 0.8) and sufficiently strong evidence [i.e., a high Bayes factor (e.g., 10(6), meaning the data are 1 million times more likely if the association is real than if there is no association)] to overcome the low prior odds for any given marker in a genome scan to be associated with a causal locus. Power calculations are given for determining the sample sizes necessary to detect effects with the required power and Bayes factor for biallelic markers in linkage disequilibrium with causal loci in additive, dominant, and recessive genetic models. Significantly stronger evidence and larger sample sizes are required than indicated by traditional hypothesis tests and power calculations. Many reported putative effects are not robustly detected and many effects including some large moderately low-frequency effects may remain undetected. These results may explain the dark matter in the genome. The power calculations have been implemented in R and will be available in the R package ldDesign.     

Linkage analysis of "necessary" disease loci versus "susceptibility" loci.

The association of some diseases with specific alleles of certain genetic markers has been difficult to explain. Several explanations have been proposed for the phenomenon of association, e.g. the existence of multiple, interacting genes (epistasis) or a disease locus in linkage disequilibrium with the marker locus. One might suppose that when marker data from families with associated diseases are analyzed for linkage, the existence of the association would assure that linkage will be found, and found at a tight recombination fraction. In fact, however, linkage analyses of some diseases associated with HLA, as well as diseases associated with alleles at other loci located throughout the genome, show significant evidence against linkage, and others show loose linkage, to the puzzlement of many researchers. In part, the puzzlement arises because linkage analysis is ideal for looking for loci that are necessary, even if not sufficient, for disease expression but may be much less useful for finding loci that are neither necessary nor sufficient for disease expression (so-called susceptibility loci). This work explores what happens when one looks for linkage to susceptibility loci. A susceptibility locus in this case means that the allele increases risk but is neither necessary nor sufficient for disease expression. It might be either an allele at the marker locus itself that is increasing susceptibility or an allele at a locus in linkage disequilibrium with the marker. This work uses computer simulation to examine how linkage analyses behave when confronted with data from such a model.(ABSTRACT TRUNCATED AT 250 WORDS)     

Evidence of QTLs on chromosomes 1q42 and 8q24 for LDL-cholesterol and apoB levels in the HERITAGE family study

Genome-wide multipoint linkage analyses were performed to identify chromosomal regions harboring genes influencing LDL-cholesterol, total apolipoprotein B (apoB), and LDL-apoB levels using 654 markers. They were assessed in a sedentary state (baseline) and after a 20 week endurance training program. Strong evidence for two quantitative trait loci (QTLs) for baseline levels was found. There is linkage evidence in black families on chromosomes 1q41-q44 [at marker D1S2860, 238 centimorgan (cM), with a maximum log of the odds (LOD) score of 3.7 for LDL-apoB] and in white families on chromosome 8q24 (at marker D8S1774, 142 cM, with LOD scores of 3.6, 3.3, and 2.5 for baseline LDL-cholesterol, LDL-apoB, and apoB, respectively). There were no strong signals for the lipoprotein training responses (as computed as the difference in posttraining minus baseline levels). In conclusion, QTLs for baseline apoB and LDL-cholesterol levels on chromosomes 1q41-q44 (in blacks) and 8q24 (in whites) were found. As there are no known strong candidate genes in these regions for lipids, follow-up studies to determine the source of those signals are needed.     

Meta-analysis of genome-wide linkage studies in BMI and obesity

Objective: The objective was to provide an overall assessment of genetic linkage data of BMI and BMI‐defined obesity using a nonparametric genome scan meta‐analysis. Research Methods and Procedures: We identified 37 published studies containing data on over 31,000 individuals from more than >10,000 families and obtained genome‐wide logarithm of the odds (LOD) scores, non‐parametric linkage (NPL) scores, or maximum likelihood scores (MLS). BMI was analyzed in a pooled set of all studies, as a subgroup of 10 studies that used BMI‐defined obesity, and for subgroups ascertained through type 2 diabetes, hypertension, or subjects of European ancestry. Results: Bins at chromosome 13q13.2‐ q33.1, 12q23‐q24.3 achieved suggestive evidence of linkage to BMI in the pooled analysis and samples ascertained for hypertension. Nominal evidence of linkage to these regions and suggestive evidence for 11q13.3‐22.3 were also observed for BMI‐defined obesity. The FTO obesity gene locus at 16q12.2 also showed nominal evidence for linkage. However, overall distribution of summed rank p values <0.05 is not different from that expected by chance. The strongest evidence was obtained in the families ascertained for hypertension at 9q31.1‐qter and 12p11.21‐q23 (p < 0.01). Conclusion: Despite having substantial statistical power, we did not unequivocally implicate specific loci for BMI or obesity. This may be because genes influencing adiposity are of very small effect, with substantial genetic heterogeneity and variable dependence on environmental factors. However, the observation that the FTO gene maps to one of the highest ranking bins for obesity is interesting and, while not a validation of this approach, indicates that other potential loci identified in this study should be investigated further.     

A genome-wide linkage scan for quantitative trait loci underlying obesity related phenotypes in 434 Caucasian families

To identify quantitative trait loci (QTLs) that contribute to obesity, we performed a large-scale whole genome linkage scan (WGS) involving 4,102 individuals from 434 Caucasian families. The most pronounced linkage evidence was found at the genomic region 20p11-12 for fat mass (LOD = 3.31) and percentage fat mass (PFM) (LOD = 2.92). We also identified several regions showing suggestive linkage signals (threshold LOD = 1.9) for obesity phenotypes, including 5q35, 8q13, 10p12, and 17q11.     

A genome-wide screen for hyposmia susceptibility Loci

Olfactory dysfunction is an important public health problem in the United States, with approximately 14 million elderly Americans having chronic olfactory impairment. We performed a genome-wide linkage scan for loci influencing susceptibility to hyposmia in the Hutterites, a founder population of European ancestry. Using interviews regarding the olfactory medical history and psychophysical smell testing, we identified 25 individuals with severe hyposmia. Elimination of subjects with confounding conditions yielded 7 hyposmics for analysis. A 52-member pedigree including all affected individuals was constructed from the larger, >1623-member pedigree, and a genome-wide screen for loci influencing the trait of hyposmia using 1123 markers was performed. The most significant evidence for linkage with hyposmia extended over a 45 cM region on chromosome 4q (P = 0.0013). Although this signal meets the criteria for suggestive linkage only and will require replication, these results offer the strongest data to date on the effects of genetic variation on olfactory dysfunction.     

A genome-wide search for genes predisposing to familial psoriasis by using a stratification approach

We have performed a genome scan, using markers spaced by 10 cM, in the search for psoriasis-susceptibility loci. The family material of 134 affected sibling pairs was ascertained on the basis of a population genetic study in which 65% of the probands had two healthy parents. Genotyping results were analyzed for non-random excessive allele-sharing between sib pairs by using GENEHUNTER ver 1.1. A stratification approach was applied to increase the homogeneity of the material by means of an operational definition of joint complaints among affected individuals. Significant linkage to the human leukocyte antigen region on chromosome 6p in a cohort including 42 families without joint complaints (nonparametric linkage score of 2.83, P=0.002) strongly supported the validity of this operational definition as it replicated results from an earlier linkage report with similar stratification criteria. New candidate regions on chromosomes 3 and 15 were identified. The highest non-parametric linkage values in this study, 2.96 (P=0.0017) and 2.89 (P=0.0020), were reached on chromosome 15 in a subgroup with joint complaints and on chromosome 3 in a subgroup without joint complaints. In addition, confirmation of previously reported loci was established on chromosomes 4q, 6p, and 17q. This study indicates that distinct disease loci might be involved in psoriasis etiology for various phenotypes.     

Genome scans provide evidence for low-HDL-C loci on chromosomes 8q23, 16q24.1-24.2, and 20q13.11 in Finnish families

We performed a genomewide scan for genes that predispose to low serum HDL cholesterol (HDL-C) in 25 well-defined Finnish families that were ascertained for familial low HDL-C and premature coronary heart disease. The potential loci for low HDL-C that were identified initially were tested in an independent sample group of 29 Finnish families that were ascertained for familial combined hyperlipidemia (FCHL), expressing low HDL-C as one component trait. The data from the previous genome scan were also reanalyzed for this trait. We found evidence for linkage between the low-HDL-C trait and three loci, in a pooled data analysis of families with low HDL-C and FCHL. The strongest statistical evidence was obtained at a locus on chromosome 8q23, with a two-point LOD score of 4.7 under a recessive mode of inheritance and a multipoint LOD score of 3.3. Evidence for linkage also emerged for loci on chromosomes 16q24.1-24.2 and 20q13.11, the latter representing a recently characterized region for type 2 diabetes. Besides these three loci, loci on chromosomes 2p and 3p showed linkage in the families with low HDL-C and a locus on 2ptel in the families with FCHL.     

Detecting gene-gene interactions using affected sib pair analysis with covariates

Interest has recently focussed on allowing for interactions between loci as a way to increase power to detect linkage. In this paper, a simplified logistic regression method was used to perform affected sib pair analyses allowing for the inclusion of data from other loci. A systematic search of two-locus disease models was carried out to determine the situations in which this was advantageous. If IBD information is available (e.g. from a genome scan), it is unlikely that allowing for interactions will give a large lod score in the absence of linkage evidence from sinlge-locus analysis. Furthermore, allowing for interactions rarely gave a significant increase in power to detect linkage over a single-locus analysis, except for heterogeneity models with low K(P). Conversely, the availability of disease-associated genotypes may greatly increase the power both to detect linkage to a second locus and interaction between the loci. These results indicate that when only IBD information is available, two-locus analysis of genome scan data should be restricted to regions giving peaks under single-locus analysis. If disease-associated genotypes are available, it may be worth re-analysing the whole genome.