Statistical Analysis For Genetic Epidemiology Research Articles

An Accidental Genetic Epidemiologist.

I briefly describe my early life and how, through a series of serendipitous events, I became a genetic epidemiologist. I discuss how the Elston-Stewart algorithm was discovered and its contribution to segregation, linkage, and association analysis. New linkage findings and paternity testing resulted from having a genotyping lab. The different meanings of interaction-statistical and biological-are clarified. The computer package S.A.G.E. (Statistical Analysis for Genetic Epidemiology), based on extensive method development over two decades, was conceived in 1986, flourished for 20 years, and is now freely available for use and further development. Finally, I describe methods to estimate and test hypotheses about familial correlations, and point out that the liability model often used to estimate disease heritability estimates the heritability of that liability, rather than of the disease itself, and so can be highly dependent on the assumed distribution of that liability.

Model-Based Linkage Analysis of a Quantitative Trait.

Linkage Analysis is a family-based method of analysis to examine whether any typed genetic markers cosegregate with a given trait, in this case a quantitative trait. If linkage exists, this is taken as evidence in support of a genetic basis for the trait. Historically, linkage analysis was performed using a binary disease trait, but has been extended to include quantitative disease measures. Quantitative traits are desirable as they provide more information than binary traits. Linkage analysis can be performed using single-marker methods (one marker at a time) or multipoint (using multiple markers simultaneously). In model-based linkage analysis the genetic model for the trait of interest is specified. There are many software options for performing linkage analysis. Here, we use the program package Statistical Analysis for Genetic Epidemiology (S.A.G.E.). S.A.G.E. was chosen because it also includes programs to perform data cleaning procedures and to generate and test genetic models for a quantitative trait, in addition to performing linkage analysis. We demonstrate in detail the process of running the program LODLINK to perform single-marker analysis, and MLOD to perform multipoint analysis using output from SEGREG, where SEGREG was used to determine the best fitting statistical model for the trait.

Model-Based Linkage Analysis of a Binary Trait.

Linkage analysis is a statistical genetics method to localize disease and trait genes to specific chromosome regions. The analysis requires pedigrees with members who vary among each other in the trait of interest and who have been genotyped with known genetic markers. Linkage analysis tests whether any of the marker alleles cosegregate with the disease or trait within the pedigree. Evidence of cosegregation is then combined across the families. We describe here the background and methods to conduct a linkage analysis for a binary trait, such as a disease, when the model of the gene contributing to the trait can be formulated. There are a number of statistical genetics software packages that allow you conduct a model-based linkage analysis of a binary trait. We describe in great detail how to run one of the programs, the LODLINK program of the Statistical Analysis for Genetic Epidemiology (S.A.G.E.) package. We provide directions for making the four input files and information on how to access and interpret the output files. We then discuss more complex analyses that can be conducted. We discuss the MLOD program for multipoint linkage analysis, including its relation to LODLINK and the additional file needed. Notes to improve your ability to run the program are included.

Model-Free Linkage Analysis of a Quantitative Trait.

Model-free methods of linkage analysis for quantitative traits are a class of easily implemented, computationally efficient and statistically robust approaches to searching for linkage to a quantitative trait. By "model-free" we refer to methods of linkage analysis that do not fully specify a genetic model (i.e., the causal allele frequency, and penetrance functions). In this chapter we briefly survey the methods that are available, and then we discuss the necessary steps to implement an analysis using the programs GENIBD, SIBPAL and RELPAL in the S.A.G.E. (Statistical Analysis for Genetic Epidemiology) software suite.

Intrafamilial Aggregation and Heritability of Aortic Reflected (Backward) Waves Derived From Wave Separation Analysis.

Although aortic wave reflection may be inherited, the extent to which indexes of wave reflection derived from wave separation analysis (reflected (backward) wave index (RI) and pressure (Pb)) show intrafamilial aggregation and heritability is uncertain. We therefore aimed to determine the intrafamilial aggregation and heritability of RI and Pb and compare these with indexes of pressure augmentation. Aortic Pb, RI, augmented pressure (Pa), and augmentation index (AIx) were determined using radial applanation tonometry and SphygmoCor software in 1,152 participants of 315 families (111 father-mother, 705 parent-child, and 301 sibling-sibling pairs) from an urban developing community of black Africans. Heritability estimates were determined from Statistical Analysis for Genetic Epidemiology software. With appropriate adjustments, significant correlations were noted between parent-child pairs for Pb and Pa (P < 0.05 for all), but not for RI (P = 0.50) or AIx (P = 0.90) and between sib-sib pairs for Pb and Pa (P < 0.05), but not for RI (P = 0.54) or AIx (P = 0.14). No correlations for indexes of wave reflection were noted between fathers and mothers (P > 0.57). After adjustments, Pb (h2 = 0.24±0.07) and Pa (h2 = 0.23±0.07) (P < 0.001 for both) but not RI (h2 = 0.04±0.06, P = 0.27) or AIx (h2 = 0.10±0.07, P = 0.07) showed significant heritability. Aortic reflected (backward) waves derived from either wave separation (Pb) or pulse wave analysis (Pa) show a similar degree of intrafamilial aggregation and heritability, but the use of RI or AIx may underestimate reflected wave effects.

Genetic epidemiological analysis of esophageal cancer in high-incidence areas of China.

Genetic epidemiological studies have shown that genetic susceptibility to esophageal cancer (EC) is an important cause of its high incidence within families in some areas of China. The purpose of this study was to obtain evidence of a genetic basis of EC in Xin-an and Xin-xiang counties in China. Familial aggregation and complex segregation analyses were performed of 79 EC families in these counties. The heritability of EC was examined using Falconer's method and complex segregation analysis was conducted with the SEGREG program in Statistical Analysis for Genetic Epidemiology (SAGE version 5.3.1). The results showed that the distribution of EC in families did not fit well into a binomial distribution. The heritability of EC among first-degree and second- degree relatives was 67.0±7.31% and 43.1%±9.80%, respectively, and the summing up powered heritability was 53.2±6.74%. The segregation ratio was 0.045. Complex segregation analysis showed that the genetic model of EC was additive. The current results provide evidence for an inherited propensity to EC in certain high-risk groups in China, and support efforts to identify the genes that confer susceptibility to this disease.

Some Capabilities for Model-Based and Model-Free Linkage Analysis using the Program Package S.A.G.E. (Statistical Analysis for Genetic Epidemiology)

For both model-free and model-based linkage analysis the S.A.G.E. (Statistical Analysis for Genetic Epidemiology) program package has some unique capabilities in analyzing both continuous traits and binary traits with variable age of onset. Here we highlight model-based linkage analysis of a quantitative trait (plasma dopamine β hydroxylase) that is known to be largely determined by monogenic inheritance, using a prior segregation analysis to produce the best fitting model for the trait. For a binary trait with variable age of onset (schizophrenia), we illustrate how using age of onset information to obtain a quantitative susceptibility trait leads to more statistically significant linkage signals, suggesting better power.

Genetic epidemiological analysis reveals a multi-gene additive model for gastric cancer

A genetic epidemiologic analysis of gastric cancer in the Chinese Han population was conducted for 64 pedigrees (902 individuals) with gastric cancer and controls obtained from the population after a census was carried out in August 2007. The heritability of gastric cancer was performed using the Falconer method and the complex segregation analysis was performed using the Statistical Analysis for Genetic Epidemiology (S.A.G.E.) SEGREG program. The heritability of gastric cancer in first- and second-degree relatives was 54.75% (95% CI, 48.01-61.49%) and 44.97% (95% CI, 33.12-56.82%), respectively. The estimated segregation ratio of gastric cancer was 0.039 (95% CI, 0.017-0.061). The complex segregation analysis showed that the Mendelian inheritance of additive model provided the best fit to the data (AIC = 170.58, P > 0.05). Therefore, polygenetic or multi-factorial additive inheritance is involved in the genetic predisposition for the development of gastric cancer.

Heritability of Susceptibility to Ionizing Radiation-Induced Apoptosis of Human Lymphocyte Subpopulations

To evaluate the heritability of intrinsic radiosensitivity, the induction of apoptosis in lymphocyte subpopulations was determined on samples from related individuals belonging to large kindred families. Quiescent lymphocytes from 334 healthy individuals were gamma-irradiated in vitro. Apoptosis was determined 18 h after irradiation by eight-color flow cytometry. Radiosensitivity was quantified from dose-effect curves. Intrafamilial correlations and heritability were computed for 199 father-mother-offspring trios using the programs SOLAR (Sequential Oligogenic Linkage Analysis Routines) and SAGE (Statistical Analysis for Genetic Epidemiology). Segregation analyses were conducted using SAGE. Marked differential susceptibility of naive and memory T lymphocytes was demonstrated. Also, although age and gender were significant covariates, their effects only accounted for a minor part of the inter-individual variation. Parent-offspring and sib-sib correlations were significant for the radiosensitivity of B cells, T4, and T8 and of effector memory T4 and T8 subpopulations. In the T4-effector memory subpopulation, the phenotype showed correlations most consistent with dominant or additive genetic effects, and the results of the segregation analysis were consistent with the contribution of a bi-allelic dominant locus. Heritability was demonstrated for the susceptibility to ionizing radiation-induced apoptosis of lymphocyte populations, and the segregation of the T4-effector memory radiosensitivity phenotype was consistent with a simple mendelian transmission model involving one major gene.

Parent-of-origin effect in the segregation analysis of bipolar affective disorder families

The purpose of this study is to test for heterogeneity in bipolar families based on the differential parental transmission of disease. Complex segregation analyses of 260 bipolar families, ascertained by the Johns Hopkins Bipolar Disorder Study, was performed based on the evidence for a parent-of-origin effect in the inheritance pattern by using REGD in Statistical Analysis for Genetic Epidemiology, Release 3.1 program. A Mendelian dominant model provided the best explanation in 57 paternal pedigrees (pedigrees with an affected paternal lineage). No evidence of Mendelian inheritance existed among 141 pedigrees showing maternal transmission. A likelihood ratio test for heterogeneity on the basis of best-fitting Mendelian dominant model showed significant differences between these two groups. These results suggest that pedigrees with no evidence of maternal transmission of bipolar disorder may represent a unique genetic subgroup of multiplex bipolar families.

Linkage and Association Analyses of Type 2 Diabetes/Impaired Glucose Metabolism and Adiponectin Serum Levels in Japanese Americans From Hawaii

Type 2 diabetes is a common disorder associated with obesity. Lower plasma levels of adiponectin were associated with type 2 diabetes. Candidate regions on chromosomes 1 ( approximately 70 cM) and 14 ( approximately 30 cM) were evaluated for replication of suggestive linkage results for type 2 diabetes/impaired glucose homeostasis in an independent sample of Japanese Americans. Replication of independent linkage evidence for serum levels of adiponectin on chromosome 14 was also evaluated. We investigated 529 subjects from 175 sibships who were originally part of the Honolulu Heart Program. Analyses included nonparametric linkage and association using SAGE (Statistical Analysis for Genetic Epidemiology) and FBAT (family-based test of association) programs and Monte Carlo simulation of Fisher's exact test in SAS. For type 2 diabetes/impaired glucose metabolism, nominal linkage evidence (P < 0.02) followed-up by genotypic association (P = 0.016) was found with marker D14S297 at 31.8 cM; linkage analyses using only diabetes phenotype were also nominally significant at this marker (P < 0.02). Nominal evidence for genotypic association to adiponectin serum level phenotype (P = 0.04) was found with the marker D14S1032 at 23.2 cM. The present study was limited by relatively small sample size. Nevertheless, these results corroborate earlier studies, suggesting that further research is warranted in the candidate region approximately 30 cM on chromosome 14.

Heritability and genetic loci of fatty liver in familial combined hyperlipidemia

VLDL overproduction, a process that is driven by an excess amount of hepatic fat, is a well-documented feature of familial combined hyperlipidemia (FCHL). The aims of this study were to investigate whether fatty liver, measured with ultrasound and as plasma alanine aminotransferase (ALT) levels, develops against a genetic background in FCHL and to identify chromosomal loci that are linked to these traits. In total, 157 FCHL family members and 20 spouses participated in this study. Radiological evidence of fatty liver was more prevalent not only in FCHL probands (40%) but also in their relatives (35%) compared with spouses (15%) (P < 0.05). Heritability calculations revealed that 20-36% of the variability in ALT levels could be attributed to genetic factors. Nonparametric quantitative trait locus (QTL) analysis revealed three significant (P < 0.001) loci with either the ultrasound or the ALT trait in the male sample: 1q42.3, 7p12-21, and 22p13-q11; none was found in the female sample or the entire group. Of these QTLs, the 7p region was consistent over time, because reanalysis with ALT levels that were determined during a visit 5 years earlier yielded similar results. This study shows that fatty liver is a heritable aspect of FCHL. Replication of particularly the 7p region is awaited.

Replication of Autism Linkage: Fine-Mapping Peak at 17q21

Autism is a heritable but genetically complex disorder characterized by deficits in language and in reciprocal social interactions, combined with repetitive and stereotypic behaviors. As with many genetically complex disorders, numerous genome scans reveal inconsistent results. A genome scan of 345 families from the Autism Genetic Resource Exchange (AGRE) (AGRE_1), gave the strongest evidence of linkage at 17q11-17q21 in families with no affected females. Here, we report a full-genome scan of an independent sample of 91 AGRE families with 109 affected sibling pairs (AGRE_2) that also shows the strongest evidence of linkage to 17q11-17q21 in families with no affected females. Taken together, these samples provide a replication of linkage to this chromosome region that is, to our knowledge, the first such replication in autism. Fine mapping at 2-centimorgan (cM) intervals in the combined sample of families with no affected females reveals a linkage peak at 66.85 cM, which places this locus at 17q21.

Inheritance of resistance to Xylella fastidiosa within a Vitis rupestris × Vitis arizonica hybrid population

The inheritance of resistance to Xylella fastidiosa (Xf), the bacterium which causes Pierce's disease (PD) in grapevines, was evaluated within a factorial mating design consisting of 16 full-sib families with resistance derived from Vitis arizonica interspecific hybrids. Measurements of disease progression under greenhouse conditions were based on quantitative assessment of Xf populations in stem tissues and on three phenotypic scores: leaf scorch, a cane maturation index (CMI) and an index that incorporated shoot stunting into the cane maturation index (CMSSI). Measurement of bacterial populations yielded the highest broad-sense heritability for resistance on a genotype mean basis (0.97), indicating that this measure of resistance was the least effected by environmental variation. Narrow-sense heritability of PD resistance was moderately high and measured 0.52, 0.60, 0.63 and 0.37 for Xf populations, CMI scores, CMSSI scores and leaf scorch values, respectively. Complex segregation analysis using the computer program Statistical Analysis for Genetic Epidemiology (SAGE: ) strongly indicated the existence of a major gene for PD resistance, which accounted for 91% of the total genetic variance. Conversion of the quantitative data into qualitative resistance levels and evaluation via a chi-square analysis showed that 15 of the 16 families segregated in accordance with a single gene hypothesis with a dominant allele controlling PD resistance. These data indicate that the trait should be relatively easy to pass on from parents to progeny in a breeding program for the development of PD-resistant grape cultivars, particularly when selection is based on cane maturation scores or stem Xf populations.

Linkage and linkage disequilibrium analysis of the lipoprotein lipase gene with lipid profiles in Chinese hypertensive families.

Elevated TG [triacylglycerol (triglyceride)] is a significant independent risk factor for cardiovascular disease. LPL (lipoprotein lipase) is one of the key enzymes in the metabolism of the TG-rich lipoproteins which hydrolyses TG from the chylomicrons and very-LDL (low-density lipoprotein). To investigate the relationship between the LPL gene and lipid profiles, especially TG, in 148 hypertensive families, we have chosen seven flanking microsatellite markers and four internal markers of the LPL gene and conducted linkage analysis by SOLAR and S.A.G.E. (statistical analysis for genetic epidemiology)/SIBPAL 2 programs, and linkage disequilibrium analysis by QTDT (quantitative transmission/disequilibrium test) and GOLD (graphical overview of linkage disequilibrium). There were statistically significant differences in lipid levels between subjects without and with hypertension within families. A maximum LOD score of 1.3 with TG at the marker D8S261 was observed by SOLAR. Using S.A.G.E./SIBPAL 2, we identified a linkage with TG at the marker 'ATTT' located within intron 6 of the LPL gene (P=0.0095). Two SNPs (single nucleotide polymorphisms), HindIII and HinfI, were found in linkage disequilibrium with LDL-cholesterol levels (P=0.0178 and P=0.0088 respectively). A strong linkage disequilibrium was observed between the HindIII in intron 8 and HinfI in the exon 9 (P<0.00001, D'=0.895). Linkage disequilibrium was also found between the 'ATTT' polymorphism in intron 6 and two SNPs (P=0.0021 and D'=0.611 for HindIII; and P=0.00004, D'=0.459 for HinfI). The present study in the Chinese families with hypertension suggested that the LPL gene might influence lipid levels, especially TG metabolism. Replication studies both in Chinese and other populations are warranted to confirm these results.

A review of the 'Statistical Analysis for Genetic Epidemiology' (SAGE) software package

The 'Statistical Analysis for Genetic Epidemiology' (S.A.G.E.) software package is an integrated, comprehensive package of computer programs designed to perform many of the different analyses required in the study of genetic epidemiology. It offers a graphical user interface for most platforms and, unlike many programs available in the public domain, is flexible in both receiving many types of input files and in allowing the user to choose among output files. All of the programs accept the same data files and together provide the means to perform familial correlation, segregation, linkage and association analyses, as well as many of the ancillary analyses that help achieve these goals. Many, but not all, of the same or similar analyses can be performed (with more difficulty) using publicly available freeware. The primary limitations of S.A.G.E. at present are the lack of software for estimating haplotypes or for identifying probable double recombinants in linkage analysis. S.A.G.E. is continually being extended and upgraded, however, with automatic downloading of the latest version always available to users.

Familial correlations and heritability of maxillary midline diastema

The purpose of the study was to estimate familial correlations and heritability to evaluate familial aggregation patterns of maxillary midline diastemas. The sample consisted of 30 extended families: 15 black, 14 white, and 1 mixed race. A single ascertainment scheme was adopted to collect the sample. Family data were collected with a 7-question survey. In all, the sample of 430 subjects consisted of 220 females, 210 males, 99 nuclear families, 534 sibling pairs, 422 avuncular pairs, 318 grandparent pairs, and 27 cousin pairs. Families were stratified by race to avoid any bias. The mixed-race family was excluded from the analysis. Data were analyzed using the program REGC in the Statistical Analysis for Genetic Epidemiology (S.A.G.E., Case Western Reserve University, Cleveland, Ohio) software. Heritability was found to be 0.32 ± 0.14 in the white sample and 0.04 ± 0.16 in the black sample. The preliminary results suggest a possible genetic basis for maxillary midline diastema and a greater role of environmental factors in the black sample than in the white sample. (Am J Orthod Dentofacial Orthop 2003;123:35-9)

Further refinement of the MYP2 locus for autosomal dominant high myopia by linkage disequilibrium analysis

Introduction: High myopia (>-6.00 diopters) is a complex common disorder that predisposes individuals to retinal detachment, glaucoma, macular degeneration, and premature cataracts. A recent linkage analysis of seven families with autosomal dominant high myopia has identified one locus (MYP2) for high myopia on chromosome 18p11.31 (Young et al.: Am J Hum Genet 1998;63:109–119). Haplotype analysis revealed an initial interval of 7.6 centimorgans (cM). Methods: Transmission disequilibrium tests (TDT) with both the Statistical Analysis for Genetic Epidemiology (SAGE) 3.1 TDTEX and GENEHUNTER 2 (GH2) programs were performed using chromosome 18p marker alleles for this interval. Results: Using SAGE analysis, the following p values were obtained for markers in marker order in this region: D18S1146 (p = 0.227), D18S481 (p = 0.001), D18S63 (p = 0.062), D18S1138 (p = 0.0004), D18S52 (p = 1.79 × 10 -6 ), and D18S62 (p = 0.141). GH2 TDT analysis revealed the following p values for the best allele for the markers: D18S1146 (p = 0.083), D18S481 (p = 0.108), D18S63 (p = 0.034), D18S1138 (p = 0.011), D18S52 (p = 0.007), and D18S62 (p = 0.479). Conclusion: These data suggest that the gene for 18p11.31-linked high myopia is most proximal to marker D18S52, with a likely interval of 0.8 cM between markers D18S63 and D18S52. Due to the contraction of the interval size by TDT, these results provide a basis for focused positional cloning and candidate gene analysis at the MYP2 locus.

No evidence of linkage between the very—low-density lipoprotein receptor gene and fasting serum insulin or homeostasis model assessment insulin resistance index: The National Heart, Lung, and Blood Institute Family Heart Study

A major gene effect on the fasting insulin level and insulin resistance has been suggested in previous studies. Several candidate genes for insulin resistance in rare syndromes have been proposed. However, there has been limited success in finding genes for common forms of insulin resistance. There is accumulating evidence of a relationship between insulin resistance and a disturbance of free fatty acid (FFA) metabolism. The very-low-density lipoprotein (VLDL) receptor, which is associated with FFA metabolism, could serve as a possible candidate gene for insulin resistance. We performed linkage analyses between the VLDL receptor gene and fasting insulin and the homeostasis model assessment (HOMA) insulin resistance index (fasting insulin x fasting glucose/22.5) in 1,050 sibpairs participating in the phase II physical examination of the National Heart, Lung, and Blood Institute Family Heart Study (FHS). Data analyses were completed using the SIBPAL component of the SAGE software package (SAGE, Statistical Analysis for Genetic Epidemiology, Version 3.1; Computer program package available from the Department of Epidemiology and Biostatistics, Case Western Reserve University, Cleveland, OH, 1997). We did not find evidence for linkage of the fasting insulin or the HOMA insulin resistance index with a polymorphic marker at the VLDL locus (P = .316 and .402, respectively). Adjustment of fasting insulin and the HOMA insulin resistance index for the body mass index (BMI) did not change the results (P = .319 and .472, respectively). In conclusion, no evidence was found for a linkage between a locus controlling the fasting insulin level or HOMA insulin resistance index and a VLDL polymorphism in the present study. Additional adjustment of fasting insulin or the HOMA insulin resistance index for the BMI did not change the linkage results significantly.

Segregation analysis of asthma: Recessive major gene component for asthma in relation to history of atopic diseases

Although asthma has a significant heritable component, the mode of inheritance remains controversial because of the complexity of the disease and the influence of environmental factors. Segregation analysis for asthma are performed with and without a history of atopic diseases (dermatitis and rhinitis) after adjusting for environmental factors. To investigate whether asthma may be inherited through a major gene with two alleles, the REGD program of the Statistical Analysis for Genetic Epidemiology (SAGE) package was conducted in 1,990 individuals from 227 families with at least one asthmatic child in a cross-sectional study of respiratory diseases in Southern Taiwan. Other covariates adjusted for included age, sex, current smoking, and environmental tobacco smoking. The hypothesis of Mendelian model and no parent-offspring transmission was rejected. However, when the variables of atopic disease and environmental factors were included in the model as covariates, the models for a two-allele gene with a recessive or codominant inheritance could not be rejected, and Akaike's Information Criterion was smaller (1,377. 13) for the recessive model than all of the other models tested, assuming a major gene with a population frequency of 0.56 +/- 0.04. However, Mendelian model without family effect was rejected. In conclusion, a history of asthma in parents is a strong risk factor for asthma in offspring. Under the assumptions of the applied segregation, at least one major gene exists that could be a gene involved also in allergy. However, the data suggest that a single locus gene explains a portion of asthma that is related to the history of atopic diseases. In addition, a polygenic/multifactorial (genetic and environmental factors) influence with a recessive component inheritance may be involved in the pathogenesis of asthma.