On the statistical properties of family-based association tests in datasets containing both pedigrees and unrelated case-control samples

Neuregulin signaling pathway in smoking behavior

Understanding molecular processes that link comorbid traits such as addictions and mental disorders can provide novel therapeutic targets. Neuregulin signaling pathway (NSP) has previously been implicated in schizophrenia, a neurodevelopmental disorder with high comorbidity to smoking. Using a Finnish twin family sample, we have previously detected association between nicotine dependence and ERBB4 (a neuregulin receptor), and linkage for smoking initiation at the ERBB4 locus on 2q33. Further, Neuregulin3 has recently been shown to associate with nicotine withdrawal in a behavioral mouse model. In this study, we scrutinized association and linkage between 15 036 common, low frequency and rare genetic variants in 10 NSP genes and phenotypes encompassing smoking and alcohol use. Using the Finnish twin family sample (N=1998 from 740 families), we detected 66 variants (representing 23 LD blocks) significantly associated (false discovery rate P<0.05) with smoking initiation, nicotine dependence and nicotine withdrawal. We comprehensively annotated the associated variants using expression (eQTL) and methylation quantitative trait loci (meQTL) analyses in a Finnish population sample. Among the 66 variants, we identified 25 eQTLs (in NRG1 and ERBB4), 22 meQTLs (in NRG3, ERBB4 and PSENEN), a missense variant in NRG1 (rs113317778) and a splicing disruption variant in ERBB4 (rs13385826). Majority of the QTLs in blood were replicated in silico using publicly available databases, with additional QTLs observed in brain. In conclusion, our results support the involvement of NSP in smoking behavior but not in alcohol use and abuse, and disclose functional potential for 56 of the 66 associated single-nucleotide polymorphism.

Association study of MMP8 gene in osteoarthritis

OBJECTIVES

Osteoarthritis (OA) is a joint disease common in the elderly. There is a prior functional evidence for different matrix metalloproteinases (MMPs), such as MMP8 and MMP9, having a role in the breakdown of cartilage extracellular matrix in OA. Thus, we analyzed whether the common genetic variants of MMP8 and MMP9 contribute to the risk of OA.

MATERIALS AND METHODS

In total, 13 common tagging single-nucleotide polymorphisms (SNPs) were studied in a discovery knee OA cohort of 185 cases and 895 controls. For validation, two knee OA replication cohorts and two hand OA replication cohorts were studied (altogether 1369 OA cases, 4445 controls in the five cohorts). The χ(2) test for individual study cohorts and Cochran-Mantel-Haenszel test for combined meta-analysis were calculated using Plink.

RESULTS

The rs1940475 SNP in MMP8 showed suggestive association in the discovery cohort (OR = 0.721, 95% CI 0.575-0.906; p = 0.005). Other knee and hand OA replication study cohorts showed similar trend for the predisposing allele without reaching statistical significance in independent replication cohorts nor in their meta-analysis (p > 0.05). Meta-analysis of all five hand and knee OA study cohorts yielded a p-value of 0.027 (OR = 0.904, 95% CI 0.826-0.989).

CONCLUSIONS

Initial analysis of the MMP8 gene showed suggestive association between rs1940475 and knee OA, but the finding did not replicate in other study cohorts, even though the trend for predisposing allele was similar in all five cohorts. MMP-8 is a good biological candidate for OA, but our study did not find common variants with significant association in the gene.

Neurocognitive performance in family-based and case-control studies of schizophrenia

BACKGROUND

Neurocognitive deficits in schizophrenia (SZ) are established and the Consortium on the Genetics of Schizophrenia (COGS) investigated such measures as endophenotypes in family-based (COGS-1) and case-control (COGS-2) studies. By requiring family participation, family-based sampling may result in samples that vary demographically and perform better on neurocognitive measures.

METHODS

The Penn computerized neurocognitive battery (CNB) evaluates accuracy and speed of performance for several domains and was administered across sites in COGS-1 and COGS-2. Most tests were included in both studies. COGS-1 included 328 patients with SZ and 497 healthy comparison subjects (HCS) and COGS-2 included 1195 patients and 1009 HCS.

RESULTS

Demographically, COGS-1 participants were younger, more educated, with more educated parents and higher estimated IQ compared to COGS-2 participants. After controlling for demographics, the two samples produced very similar performance profiles compared to their respective controls. As expected, performance was better and with smaller effect sizes compared to controls in COGS-1 relative to COGS-2. Better performance was most pronounced for spatial processing while emotion identification had large effect sizes for both accuracy and speed in both samples. Performance was positively correlated with functioning and negatively with negative and positive symptoms in both samples, but correlations were attenuated in COGS-2, especially with positive symptoms.

CONCLUSIONS

Patients ascertained through family-based design have more favorable demographics and better performance on some neurocognitive domains. Thus, studies that use case-control ascertainment may tap into populations with more severe forms of illness that are exposed to less favorable factors compared to those ascertained with family-based designs.

PSEUDOMARKER 2.0: efficient computation of likelihoods using NOMAD

BACKGROUND

PSEUDOMARKER is a software package that performs joint linkage and linkage disequilibrium analysis between a marker and a putative disease locus. A key feature of PSEUDOMARKER is that it can combine case-controls and pedigrees of varying structure into a single unified analysis. Thus it maximizes the full likelihood of the data over marker allele frequencies or conditional allele frequencies on disease and recombination fraction.

RESULTS

The new version 2.0 uses the software package NOMAD to maximize likelihoods, resulting in generally comparable or better optima with many fewer evaluations of the likelihood functions.

CONCLUSIONS

After being modified substantially to use modern optimization methods, PSEUDOMARKER version 2.0 is more robust and substantially faster than version 1.0. NOMAD may be useful in other bioinformatics problems where complex likelihood functions are optimized.

Identifying flavor preference subgroups. Genetic basis and related eating behavior traits

Subgroups based on flavor preferences were identified and their genetic and behavior related characteristics investigated using extensive data from 331 Finnish twins (21-25years, 146 men) including 47 monozygotic (MZ) and 93 dizygotic (DZ) pairs, and 51 twin individuals. The subgroup identification (hierarchical and K-means clustering) was based on liking responses to food names representing sour, umami, and spicy flavor qualities. Furthermore, sensory tests were conducted, a questionnaire on food likes completed, and various eating behavior related traits measured with validated scales. Sensory data included intensity ratings of PROP (6-n-propylthiouracil-impregnated filter paper), hedonic and intensity responses to sourness (orange juice with and without added citric acid, 0.42%), pungency (strawberry jelly with and without added capsaicin 0.00013%) and umami ('mouthfeel flavor' taste solution). Ratings of liking of 41 general food names were categorized into salty-and-fatty, sweet-and-fatty, fruits and vegetables and fish foods. Subgroup differences (complex samples procedure) and the genetics underlying the subgroups (structural equation modeling) were investigated. Of the resulting two groups (basic, n=140, adventurous n=152; non-grouped n=39), the adventurous expressed higher liking for sour and spicy foods, and had more tolerance for capsaicin burn in the sensory-hedonic test. The adventurous were also less food neophobic (25.9±9.1 vs. 32.5±10.6, respectively) and expressed higher liking for fruits and vegetables compared to the basic group. Genetic effects were shown to underlie the subgroups (heritability 72%, CI: 36-92%). Linkage analysis for 27 candidate gene regions revealed suggestively that being adventurous is linked to TAS1R1 and PKD1L3 genes. These results indicate that food neophobia and genetic differences may form a barrier through which individual flavor preferences are generated.

Detection and impact of rare regulatory variants in human disease

Advances in genome sequencing are providing unprecedented resolution of rare and private variants. However, methods which assess the effect of these variants have relied predominantly on information within coding sequences. Assessing their impact in non-coding sequences remains a significant contemporary challenge. In this review, we highlight the role of regulatory variation as causative agents and modifiers of monogenic disorders. We further discuss how advances in functional genomics are now providing new opportunity to assess the impact of rare non-coding variants and their role in disease.

A large-scale candidate gene analysis of mood disorders: evidence of neurotrophic tyrosine kinase receptor and opioid receptor signaling dysfunction

BACKGROUND

Despite proven heritability, little is known about the genetic architecture of mood disorders. Although a number of family and case-control studies have examined the genetics of mood disorders, none have carried out joint linkage-association studies and sought to validate the results with gene expression analyses in an independent cohort.

METHODS

We present findings from a large candidate gene study that combines linkage and association analyses using families and singletons, providing a systematic candidate gene investigation of mood disorder. For this study, 876 individuals were recruited, including 83 families with 313 individuals and 563 singletons. This large-scale candidate gene analysis included 130 candidate genes implicated in addictive and other psychiatric disorders. These data showed significant genetic associations for 28 of these candidate genes, although none remained significant after correction for multiple testing. To evaluate the functional significance of these 28 candidate genes in mood disorders, we examined the transcriptional profiles of these genes within the dorsolateral prefrontal cortex and anterior cingulate for 21 cases with mood disorders and 25 nonpsychiatric controls, and carried out a pathway analysis to identify points of high connectivity suggestive of particular molecular pathways that may be dysregulated.

RESULTS

Two primary gene candidates were supported by the linkage-association, gene expression profiling, and network analysis: neurotrophic tyrosine kinase receptor, type 2 (NTRK2), and the opioid receptor, κ1 (OPRK1).

CONCLUSION

This study supports a role for NTRK2 and OPRK1 signaling in the pathophysiology of mood disorder. The unique approach incorporating evidence from multiple experimental and computational modalities enhances confidence in these findings.

Family-based association study between SLC2A1, HK1, and LEPR polymorphisms with myelomeningocele in Chile

Obese/diabetic mothers present a higher risk to develop offspring with myelomeningocele (MM), evidence supporting the role of energy homeostasis-related genes in neural tube defects. Using polymerase chain reaction-restriction fragment length polymorphism, we have genotyped SLC2A1, HK1, and LEPR single-nucleotide polymorphisms in 105 Chilean patients with MM and their parents in order to evaluate allele-phenotype associations by means of allele/haplotype transmission test (TDT) and parent-of-origin effects. We detected an undertransmission for the SLC2A1 haplotype T-A (rs710218-rs2229682; P = .040), which was not significant when only lower MM (90% of the cases) was analyzed. In addition, the leptin receptor rs1137100 G allele showed a significant increase in the risk of MM for maternal-derived alleles in the whole sample (2.43-fold; P = .038) and in lower MM (3.20-fold; P = .014). Our results support the role of genes involved in energy homeostasis in the risk of developing MM, thus sustaining the hypothesis of diverse pathways and genetic mechanisms acting in the expression of such birth defect.

Heritability of ambulatory and beat-to-beat office blood pressure in large multigenerational Arab pedigrees: the 'Oman Family study'

OBJECTIVE

To estimate the heritability of ambulatory blood pressure (BP), heart rate (HR), and beat-to-beat office BP and HR in an isolated, environmentally and genetically homogeneous Omani Arab population.

METHODS

Ambulatory BP measurements were recorded in 1,124 subjects with a mean age of 33.8 ± 16.2 years, using the auscultatory mode of the validated Schiller ambulatory BP Monitor. Beat-to-beat BP and HR were recorded by the Task Force Monitor. Heritability was estimated using quantitative genetic analysis. This was achieved by applying the maximum-likelihood-based variance decomposition method implemented in SOLAR software.

RESULTS

We detected statistically significant heritability estimates for office beat-to-beat, 24-hour, daytime, and sleep HR of 0.31, 0.21, 0.20, and 0.07, respectively. Heritability estimates in the above mentioned conditions for systolic BP (SBP)/diastolic BP (DBP)/mean BP (MBP)were all significant and estimated at 0.19/0.19/0.19, 0.30/0.44/0.41, 0.28/0.38/0.39, and 0.21/0.18/0.20,respectively. Heritability estimates for 24-hour and daytime ambulatory SBP, DBP, and MBP ranged from 0.28 to 0.44, and were higher than the heritability estimates for beat-to-beat recordings and sleep periods,which were estimated within a narrow range of 0.18-0.21.

CONCLUSION

In this cohort, because shared environments are common to all, the environmental influence that occurs is primarily due to the variation in non-shared environment that is unique to the individual. We demonstrated significant heritability estimates for both beat-to-beat office and ambulatory BP and HR recordings, but 24-hour and daytime ambulatory heritabilities are higher than those from beat-to-beat resting levels and ambulatory night-time recordings.

On the validity of the likelihood ratio test and consistency of resulting parameter estimates in joint linkage and linkage disequilibrium analysis under improperly specified parametric models

It has been shown that parametric analysis of linkage disequilibrium conditional on linkage using an overly deterministic model can be optimal for family-based association analysis. However, if one applies this strategy carelessly, there is a risk of false inference. We analyse properties of such likelihood ratio tests when the assumed disease mode of inheritance is inaccurate. Under some conditions, problems result if one is not careful to consider what null hypothesis is being tested. We show that: (a) tests for which the null hypothesis assumes the absence of both linkage and association are independent of the true mode of inheritance; (b) likelihood ratio tests assuming either linkage or association under the null hypothesis may depend on the true mode of inheritance, leading to inconsistent parameter estimates, in particular under extremely deterministic models; (c) this problem cannot be eliminated by increasing sample size or adding population controls--as sample size increases, the chance of false positive inference goes to 100%; (d) this issue can lead to systematic false positive inference of association in regions of linkage. This is important because highly deterministic models are often used intentionally in model-based analyses because they can have more power than the true model, and are implicit in many model-free analysis methods.

PSEUDOMARKER: a powerful program for joint linkage and/or linkage disequilibrium analysis on mixtures of singletons and related individuals

A decade ago, there was widespread enthusiasm for the prospects of genome-wide association studies to identify common variants related to common chronic diseases using samples of unrelated individuals from populations. Although technological advancements allow us to query more than a million SNPs across the genome at low cost, a disappointingly small fraction of the genetic portion of common disease etiology has been uncovered. This has led to the hypothesis that less frequent variants might be involved, stimulating a renaissance of the traditional approach of seeking genes using multiplex families from less diverse populations. However, by using the modern genotyping and sequencing technology, we can now look not just at linkage, but jointly at linkage and linkage disequilibrium (LD) in such samples. Software methods that can look simultaneously at linkage and LD in a powerful and robust manner have been lacking. Most algorithms cannot jointly analyze datasets involving families of varying structures in a statistically or computationally efficient manner. We have implemented previously proposed statistical algorithms in a user-friendly software package, PSEUDOMARKER. This paper is an announcement of this software package. We describe the motivation behind the approach, the statistical methods, and software, and we briefly demonstrate PSEUDOMARKER's advantages over other packages by example.