Machine learning approaches to explore digenic inheritance.

Some rare genetic disorders, such as retinitis pigmentosa or Alport syndrome, are caused by the co-inheritance of DNA variants at two different genetic loci (digenic inheritance). To capture the effects of these disease-causing variants and their possible interactive effects, various statistical methods have been developed in human genetics. Analogous developments have taken place in the field of machine learning, particularly for the field that is now called Big Data. In the past, these two areas have grown independently and have started to converge only in recent years. We discuss an overview of each of the two fields, paying special attention to machine learning methods for uncovering the combined effects of pairs of variants on human disease.

Implication of Melanocortin Receptor Genes in the Familial Comorbidity of Type 2 Diabetes and Depression.

The melanocortin receptors are G-protein-coupled receptors, which are essential components of the hypothalamic-pituitary-adrenal axis, and they mediate the actions of melanocortins (melanocyte-stimulating hormones: α-MSH, ß-MSH, and γ-MSH) as well as the adrenocorticotropin hormone (ACTH) in skin pigmentation, adrenal steroidogenesis, and stress response. Three melanocortin receptor genes (MC1R, MC2R, and MC5R) contribute to the risk of major depressive disorder (MDD), and one melanocortin receptor gene (MC4R) contributes to the risk of type 2 diabetes (T2D). MDD increases T2D risk in drug-naïve patients; thus, MDD and T2D commonly coexist. The five melanocortin receptor genes might confer risk for both disorders. However, they have never been investigated jointly to evaluate their potential contributing roles in the MDD-T2D comorbidity, specifically within families. In 212 Italian families with T2D and MDD, we tested 11 single nucleotide polymorphisms (SNPs) in the MC1R gene, 9 SNPs in MC2R, 3 SNPs in MC3R, 4 SNPs in MC4R, and 2 SNPs in MC5R. The testing used 2-point parametric linkage and linkage disequilibrium (LD) (i.e., association) analysis with four models (dominant with complete penetrance (D1), dominant with incomplete penetrance (D2), recessive with complete penetrance (R1), and recessive with incomplete penetrance (R2)). We detected significant (p ≤ 0.05) linkage and/or LD (i.e., association) to/with MDD for one SNP in MC2R (rs111734014) and one SNP in MC5R (rs2236700), and to/with T2D for three SNPs in MC1R (rs1805007 and rs201192930, and rs2228479), one SNP in MC2R (rs104894660), two SNPs in MC3R (rs3746619 and rs3827103), and one SNP in MC4R genes (Chr18-60372302). The linkage/LD/association was significant across different linkage patterns and different modes of inheritance. All reported variants are novel in MDD and T2D. This is the first study to report risk variants in MC1R, MC2R, and MC3R genes in T2D. MC2R and MC5R genes are replicated in MDD, with one novel variant each. Within our dataset, only the MC2R gene appears to confer risk for both MDD and T2D, albeit with different risk variants. To further clarity the role of the melanocortin receptor genes in MDD-T2D, these findings should be sought among other ethnicities as well.

Comorbidity of Novel CRHR2 Gene Variants in Type 2 Diabetes and Depression.

The corticotropin-releasing hormone receptor 2 (CRHR2) gene encodes CRHR2, contributing to the hypothalamic-pituitary-adrenal stress response and to hyperglycemia and insulin resistance. CRHR2-/- mice are hypersensitive to stress, and the CRHR2 locus has been linked to type 2 diabetes and depression. While CRHR2 variants confer risk for mood disorders, MDD, and type 2 diabetes, they have not been investigated in familial T2D and MDD. In 212 Italian families with type 2 diabetes and depression, we tested 17 CRHR2 single nucleotide polymorphisms (SNPs), using two-point parametric-linkage and linkage-disequilibrium (i.e., association) analysis (models: dominant-complete-penetrance-D1, dominant-incomplete-penetrance-D2, recessive-complete-penetrance-R1, recessive-incomplete-penetrance-R2). We detected novel linkage/linkage-disequilibrium/association to/with depression (3 SNPs/D1, 2 SNPs/D2, 3 SNPs/R1, 3 SNPs/R2) and type 2 diabetes (3 SNPs/D1, 2 SNPs/D2, 2 SNPs/R1, 1 SNP/R2). All detected risk variants are novel. Two depression-risk variants within one linkage-disequilibrium block replicate each other. Two independent novel SNPs were comorbid while the most significant conferred either depression- or type 2 diabetes-risk. Although the families were primarily ascertained for type 2 diabetes, depression-risk variants showed higher significance than type 2 diabetes-risk variants, implying CRHR2 has a stronger role in depression-risk than type 2 diabetes-risk. In silico analysis predicted variants' dysfunction. CRHR2 is for the first time linked to/in linkage-disequilibrium/association with depression-type 2 diabetes comorbidity and may underlie the shared genetic pathogenesis via pleiotropy.

Shared genomic segment analysis with equivalence testing.

An important aspect of disease gene mapping is replication, that is, a putative finding in one group of individuals is confirmed in another set of individuals. As it can happen by chance that individuals share an estimated disease position, we developed a statistical approach to determine the p-value for multiple individuals or families to share a possibly small number of candidate susceptibility variants. Here, we focus on candidate variants for dominant traits that have been obtained by our previously developed heterozygosity analysis, and we are testing the sharing of candidate variants obtained for different individuals. Our approach allows for multiple pathogenic variants in a gene to contribute to disease, and for estimated disease variant positions to be imprecise. Statistically, the method developed here falls into the category of equivalence testing, where the classical null and alternative hypotheses of homogeneity and heterogeneity are reversed. The null hypothesis situation is created by permuting genomic locations of variants for one individual after another. We applied our methodology to the ALSPAC data set of 1,927 whole-genome sequenced individuals, where some individuals carry a pathogenic variant for the BRCA1 gene, but no two individuals carry the same variant. Our shared genomic segment analysis found significant evidence for BRCA1 pathogenic variants within ±5 kb of a given DNA variant.

Population genetics: past, present, and future.

We present selected topics of population genetics and molecular phylogeny. As several excellent review articles have been published and generally focus on European and American scientists, here, we emphasize contributions by Japanese researchers. Our review may also be seen as a belated 50-year celebration of Motoo Kimura's early seminal paper on the molecular clock, published in 1968.

Genetic linkage analysis in the age of whole-genome sequencing.

For many years, linkage analysis was the primary tool used for the genetic mapping of Mendelian and complex traits with familial aggregation. Linkage analysis was largely supplanted by the wide adoption of genome-wide association studies (GWASs). However, with the recent increased use of whole-genome sequencing (WGS), linkage analysis is again emerging as an important and powerful analysis method for the identification of genes involved in disease aetiology, often in conjunction with WGS filtering approaches. Here, we review the principles of linkage analysis and provide practical guidelines for carrying out linkage studies using WGS data.

Polymorphisms in Stress-Related Genes Are Associated with Reduced Cocaine Abuse and Longer Retention in Methadone Maintenance Treatment for Opioid Use Disorder.

BACKGROUND: As CRH-binding protein (CRHBP) SNP rs1500 was associated with reduced cocaine abuse after 1 year in methadone maintenance treatment (MMT) for heroin addiction, we evaluated the association of additional 28 selected SNPs, in 17 stress-related genes, with MMT outcome. METHODS: The distribution of genotypes of each SNP by cocaine abuse after 1 year in MMT was assessed under the dominant and recessive models using χ2. Cumulative retention (up to 26.5 years) was studied using Kaplan-Meier analyses. Logistic regression and Cox model were used for multivariate analyses. RESULTS: Of a nonselective sample of 404 patients, 25 patients with <50% Europeans/Middle Eastern ancestry were excluded. Of the remaining 379 patients, 330 (87.1%) stayed at least 1 year in treatment. Four SNPs were associated with cocaine abuse after 1 year in MMT. A lower proportion of cocaine abusers was found in the groups of subjects with the following genotypes: arginine vasopressin (AVP) SNP rs2282018 CC, CRHBP rs7728378 TT, galanin rs3136541 TT/TC, and neuropeptide Y receptor Y1 (NPY1R) rs4518200 AA. The following independent variables were associated with lack of cocaine in urine after 1 year (multivariate analyses): CRHBP rs7728378 TT, NPY1R rs4518200 AA, no cocaine in urine on admission, as well as opiate and benzodiazepine use after 1 year in MMT. Cumulative retention (n = 379) was longer in carriers of AVP rs2282018 CC (13.7 years, 95% CI 11.1-16.2) versus TT/TC genotypes (10.5, 95% CI 9.4-11.5) (p = 0.0230) Conclusions: The study suggests that a reduction in cocaine abuse and longer retention among MMT patients is mediated in part by variants in stress-related genes and is a step toward precision medicine.

A novel association of rs13334070 in the RPGRIP1L gene with adiposity factors discovered by joint linkage and linkage disequilibrium analysis in Iranian pedigrees: Tehran Cardiometabolic Genetic Study (TCGS).

Understanding the genetic and metabolic bases of obesity is helpful in planning and developing health strategies. Therefore, the first family-based joint linkage and linkage disequilibrium study was conducted in Iranian pedigrees to assess the relationship between obesity and single-nucleotide polymorphisms (SNPs) located in the 16q12.2 region. In the present study, a total of 13,344 individuals were included, of whom 12,502 individuals were within 3,109 pedigrees and 842 were unrelated singletons. To investigate the relationship between obesity and genetic variants, a joint model of linkage and linkage disequilibrium was applied. Moreover, a sequence kernel association test (SKAT) was used to evaluate the association of the SNP set with body size and lipid profile measurements. The joint model showed that rs13334070, in the intron 4 of the RPGRIP1L gene, has a significant association with obesity. According to the 4-gamete rule, which is a procedure for constructing SNP sets by considering recombination occurrence between SNPs, this polymorphism has a high correlation with six nearby SNPs that make an SNP set. SKAT showed that this SNP set has a significant association with body size factors, but almost no association with most of the lipid profile measurements. In conclusion, from the result of this study, it might be reasonable to consider RPGRIP1L as an important gene whose variations could be associated with obesity risk factors.

Variants of opioid genes and response to treatment of opioid use disorder with buprenorphine-naloxone versus extended-release naltrexone in Caucasians.

Background: Sublingual buprenorphine-naloxone (BUP-NX), an FDA-approved treatment for opioid use disorder (OUD), combines buprenorphine (a partial mu/kappa agonist) with naloxone (a mu/ kappa antagonist). Extended-release injection naltrexone (XR-NTX; a mu receptor antagonist and kappa receptor partial agonist) is also an FDA-approved treatment for OUD. However, while some patients respond well to these medications, many others leave treatment and relapse. Objectives: Determine whether gene variants in the opioid gene system are associated with better or worse treatment response. Methods: In a 24-week, multisite, randomized, comparative effectiveness trial of daily, sublingual self-administration of BUP-NX versus monthly injection of XR-NTX conducted in the National Drug Abuse Clinical Trials Network, DNA was collected and four opioid gene variants were evaluated: (1) mu opioid receptor 118A>G; (2) 68-bp repeat in prodynorphin; (3) prodynorphin SNP rs910080; and (4) kappa opioid receptor SNP rs6473797. In non-Hispanic Caucasians (N = 334), two outcomes measures were assessed: received first dose (yes/no) and received last dose (yes/no). Separate logistic regressions were used to model each outcome measure as a function of treatment (XR-NTX vs BUP-NX), each gene variant, and their interaction. Results: There were no significant main effects of gene variant on receiving first dose or last dose. There were also no significant gene variant by treatment interactions. Conclusions: The outcome of treatment of OUD with medications is likely a complex function of multiple factors, including environmental, psychosocial, and possibly genetic, such that major effects of genetic variants may be unlikely.

A Novel Locus and Candidate Gene for Familial Developmental Dyslexia on Chromosome 4q.

Objective: Developmental dyslexia is a highly heritable specific reading and writing disability. To identify a possible new locus and candidate gene for this disability, we investigated a four-generation pedigree where transmission of dyslexia is consistent with an autosomal dominant inheritance pattern. Methods: We performed genome wide array-based SNP genotyping and parametric linkage analysis and sequencing analysis of protein-coding exons, exon-intron boundaries and conserved extragenic regions within the haplotype cosegregating with dyslexia in DNA from one affected and one unaffected family member. Cosegregation was confirmed by sequencing all available family members. Additionally, we analyzed 96 dyslexic individuals who had previously shown positive LOD scores on chromosome 4q28 as well as an even larger sample (n = 2591). Results: We found a single prominent linkage interval on chromosome 4q, where sequence analysis revealed a nucleotide variant in the 3' UTR of brain expressed SPRY1 in the dyslexic family member that cosegregated with dyslexia. This sequence alteration might affect the binding efficiency of the IGF2BP1 RNA-binding protein and thus influence the expression level of the SPRY1 gene product. An analysis of 96 individuals from a cohort of dyslexic individuals revealed a second heterozygous variant in this gene, which was absent in the unaffected sister of the proband. An investigation of the region in a much larger sample further found a nominal p-value of 0.0016 for verbal short-term memory (digit span) in 2,591 individuals for a neighboring SNV. After correcting for the local number of analyzed SNVs, and after taking into account linkage disequilibrium, we found this corresponds to a p-value of 0.0678 for this phenotype. Conclusions: We describe a new locus for familial dyslexia and discuss the possibility that SPRY1 might play a role in the etiology of a monogenic form of dyslexia.

Heterozygosity mapping for human dominant trait variants.

Homozygosity mapping is a well-known technique to identify runs of homozygous variants that are likely to harbor genes responsible for autosomal recessive disease, but a comparable method for autosomal dominant traits has been lacking. We developed an approach to map dominant disease genes based on heterozygosity frequencies of sequence variants in the immediate vicinity of a dominant trait. We demonstrate through theoretical analysis that DNA variants surrounding an inherited dominant disease variant tend to have increased heterozygosity compared with variants elsewhere in the genome. We confirm existence of this phenomenon in sequence data with known dominant pathogenic variants obtained on family members and in unrelated population controls. A computer-based approach to estimating empirical significance levels associated with our test statistics shows genome-wide p-values smaller than 0.05 for many but not all of the individuals carrying a pathogenic variant.

Dysfunctional nitric oxide signalling increases risk of myocardial infarction.

Myocardial infarction, a leading cause of death in the Western world, usually occurs when the fibrous cap overlying an atherosclerotic plaque in a coronary artery ruptures. The resulting exposure of blood to the atherosclerotic material then triggers thrombus formation, which occludes the artery. The importance of genetic predisposition to coronary artery disease and myocardial infarction is best documented by the predictive value of a positive family history. Next-generation sequencing in families with several affected individuals has revolutionized mutation identification. Here we report the segregation of two private, heterozygous mutations in two functionally related genes, GUCY1A3 (p.Leu163Phefs*24) and CCT7 (p.Ser525Leu), in an extended myocardial infarction family. GUCY1A3 encodes the α1 subunit of soluble guanylyl cyclase (α1-sGC), and CCT7 encodes CCTη, a member of the tailless complex polypeptide 1 ring complex, which, among other functions, stabilizes soluble guanylyl cyclase. After stimulation with nitric oxide, soluble guanylyl cyclase generates cGMP, which induces vasodilation and inhibits platelet activation. We demonstrate in vitro that mutations in both GUCY1A3 and CCT7 severely reduce α1-sGC as well as ß1-sGC protein content, and impair soluble guanylyl cyclase activity. Moreover, platelets from digenic mutation carriers contained less soluble guanylyl cyclase protein and consequently displayed reduced nitric-oxide-induced cGMP formation. Mice deficient in α1-sGC protein displayed accelerated thrombus formation in the microcirculation after local trauma. Starting with a severely affected family, we have identified a link between impaired soluble-guanylyl-cyclase-dependent nitric oxide signalling and myocardial infarction risk, possibly through accelerated thrombus formation. Reversing this defect may provide a new therapeutic target for reducing the risk of myocardial infarction.

HDR-del: A tool based on Hamming distance for prioritizing pathogenic chromosomal deletions in exome sequencing.

High-density oligonucleotide arrays have widely been used to detect pathogenic chromosomal deletions. In addition to high-density oligonucleotide arrays, programs using whole-exome sequencing have become available for estimating copy-number variations using depth of coverage. Here, we propose a new statistical method, HDR-del, to prioritize pathogenic chromosomal deletions based on Hamming distance in exome sequencing. In vcf (variant call format) files generated from exome sequencing, hemizygous chromosomal deletion regions lack heterozygous variants and lead to apparent long runs of homozygosity (ROH). In our Hamming distance ratio (HDR)-del approach, we calculate the "difference" in heterozygous status between an affected individual and control individuals using the HDR over all candidate chromosomal deletion regions defined as ROH longer than 1Mbp. Using a suitable test statistic, which is expected to be large for a true pathogenic deletion region, we prioritize candidate chromosomal deletion regions based on this statistic. In our approach, we were able to considerably narrow down true pathogenic chromosomal deletion regions, which were confirmed by high-density oligonucleotide arrays in four mitochondrial disease patients. Our HDR-del approach represents an easy method for detecting chromosomal deletions.

Family-based designs for genome-wide association studies.

Association mapping has successfully identified common SNPs associated with many diseases. However, the inability of this class of variation to account for most of the supposed heritability has led to a renewed interest in methods - primarily linkage analysis - to detect rare variants. Family designs allow for control of population stratification, investigations of questions such as parent-of-origin effects and other applications that are imperfectly or not readily addressed in case-control association studies. This article guides readers through the interface between linkage and association analysis, reviews the new methodologies and provides useful guidelines for applications. Just as effective SNP-genotyping tools helped to realize the potential of association studies, next-generation sequencing tools will benefit genetic studies by improving the power of family-based approaches.

HDR: a statistical two-step approach successfully identifies disease genes in autosomal recessive families.

In the search for sequence variants underlying disease, commonly applied filtering steps usually result in a number of candidate variants that cannot further be narrowed down. In autosomal recessive families, disease usually occurs only in one generation so that genetic linkage analysis is unlikely to help. Because homozygous recessive mutations tend to be inherited together with flanking homozygous variants, we developed a statistical method to detect pathogenic variants in autosomal recessive families: We look for differences in patterns of homozygosity around candidate variants between patients and control individuals and expect that such differences are greater for pathogenic variants than random candidate variants. In six autosomal recessive mitochondrial disease families, in which pathogenic homozygous variants have already been identified, our approach succeeded in prioritizing pathogenic mutations. Our method is applicable to single patients from recessive families with at least a few dozen control individuals from the same population; it is easy to use and is highly effective for detecting causative mutations in autosomal recessive families.

Polygenic Models for Risk Prediction in Human Genetics.

After a brief discussion and evaluation of Fisher's additive polygenic inheritance model, a few other approaches to polygenic inheritance are discussed, notably the polygenic threshold model. A literature review of applications of such models in human genetics is presented, and a simple approach, based on a finite polygenic model, is developed in detail. The fit of the model to recurrence risks in schizophrenia is discussed.

Genome-wide association scan in north Indians reveals three novel HLA-independent risk loci for ulcerative colitis.

OBJECTIVE: Over 100 ulcerative colitis (UC) loci have been identified by genome-wide association studies (GWASs) primarily in Caucasians (CEUs). Many of them have weak effects on disease susceptibility, and the bulk of the heritability cannot be ascribed to these loci. Very little is known about the genetic background of UC in non-CEU groups. Here we report the first GWAS on UC in a genetically distinct north Indian (NI) population. DESIGN: A genome-wide scan was performed on 700 cases and 761 controls. 18 single-nucleotide polymorphisms (SNPs) (p<5×10(-5)) were genotyped in an independent cohort of 733 cases and 1148 controls. A linear mixed model was used for case-control association tests. RESULTS: Seven novel human leucocyte antigen (HLA)-independent SNPs from chromosome 6, located in 3.8-1, BAT2, MSH5, HSPA1L, SLC44A4, CFB and NOTCH4, exceeded p<5×10(-8) in the combined analysis. To assess the independent biological contribution of such genes from the extended HLA region, we determined the percentage alternative pathway activity of complement factor B (CFB), the top novel hit. The activity was significantly different (p=0.01) between the different genotypes at rs12614 in UC cases. Transethnic comparisons revealed a shared contribution of a fraction of UC risk genes between NI and CEU populations, in addition to genetic heterogeneity. CONCLUSIONS: This study shows varying contribution of the HLA region to UC in different populations. Different environmental exposures and the characteristic genetic structure of the HLA locus across ethnic groups collectively make it amenable to the discovery of causative alleles by transethnic resequencing. This may lead to an improved understanding of the molecular mechanisms underlying UC.

Common Regulatory Variants of CYFIP1 Contribute to Susceptibility for Autism Spectrum Disorder (ASD) and Classical Autism.

Based on the analysis of mRNA expression and genotype data from the "Brain Cloud" database, we identified seven SNPs within or near the autism candidate gene CYFIP1 that show nominally significant correlations between genotype and CYFIP1 mRNA expression in human dorsolateral prefrontal cortex. Analysis of transmission disequilibrium test (TDT) odds ratios (ORs) for these SNPs in a large Autism Genome Project (AGP) trio-based association study revealed the high-expression alleles of four of these SNPs (rs8028440, rs2289823, rs7403800 and rs3751566) to be susceptibility alleles. Correlations between the regression coefficients for mRNA expression and log10 -transformed TDT ORs were statistically significant [P = 0.008 (ASD); P = 0.002 (classical autism)]. Similarly, statistically significant correlations were obtained between levels of CYFIP1 mRNA expression predicted using the regression equations obtained from multiple linear regression analysis and log10 -transformed TDT ORs for specific combinations of genotypes for both ASD (rs2289823 + rs3751566: P = 0.008) and classical autism (rs2289823 + rs3751566: P = 0.008; rs2289823 + rs3751566 + rs765763: P = 0.0006) diagnoses. Together, these results support the hypothesis that high expression of CYFIP1 mRNA increases susceptibility for both ASD and classical autism.

Leveling the Playing Field in Homozygosity Mapping Using Map Distances.

Studies of linkage disequilibrium (LD) and its variation in the genome are of central importance for understanding evolutionary history, population structure, and selective sweeps. Extreme forms of the latter may result in runs of homozygosity (ROH). In human gene mapping, long ROHs are the basis for homozygosity mapping (HM) with length measured in terms of Mb (106 base pairs physical distance). LD varies greatly over the human genome so that long ROHs tend to occur preferentially in regions of high LD and ROHs of the same length in different regions are not strictly comparable. Thus, in human gene mapping, LD appears as a confounder that needs to be taken into account in the interpretation of ROHs. The effect of varying LD can be mitigated by working on a scale of centimorgans (cM, genetic distance) instead of Mb. We demonstrate this effect for HapMap 3 data on chromosome 19 and show examples with different ROH lengths depending on whether physical or genetic lengths are used. These results suggest that HM should preferably be done on genetic rather than physical distances.

Overlapping dopaminergic pathway genetic susceptibility to heroin and cocaine addictions in African Americans.

Drugs of abuse activate the mesolimbic dopaminergic pathway. Genetic variations in the dopaminergic system may contribute to drug addiction. Several processes are shared between cocaine and heroin addictions but some neurobiological mechanisms may be specific. This study examined the association of 98 single nucleotide polymorphisms in 13 dopamine-related genes with heroin addiction (OD) and/or cocaine addiction (CD) in a sample of 801 African Americans (315 subjects with OD ± CD, 279 subjects with CD, and 207 controls). Single-marker analyses provided nominally significant evidence for associations of 24 SNPs) in DRD1, ANKK1/DRD2, DRD3, DRD5, DBH, DDC, COMT and CSNK1E. A DRD2 7-SNPs haplotype that includes SNPs rs1075650 and rs2283265, which were shown to alter D2S/D2L splicing, was indicated in both addictions. The Met allele of the functional COMT Val158Met was associated with protection from OD. None of the signals remained significant after correction for multiple testing. The study results are in accordance with the results of previous studies, including our report of association of DRD1 SNP rs5326 with OD. The findings suggest the presence of an overlap in genetic susceptibility for OD and CD, as well as shared and distinct susceptibility for OD in subjects of African and European descent.