Machine Learning for Detecting Gene-Gene Interactions

Complex interactions among genes and environmental factors are known to play a role in common human disease aetiology. There is a growing body of evidence to suggest that complex interactions are 'the norm' and, rather than amounting to a small perturbation to classical Mendelian genetics, interactions may be the predominant effect. Traditional statistical methods are not well suited for detecting such interactions, especially when the data are high dimensional (many attributes or independent variables) or when interactions occur between more than two polymorphisms. In this review, we discuss machine-learning models and algorithms for identifying and characterising susceptibility genes in common, complex, multifactorial human diseases. We focus on the following machine-learning methods that have been used to detect gene-gene interactions: neural networks, cellular automata, random forests, and multifactor dimensionality reduction. We conclude with some ideas about how these methods and others can be integrated into a comprehensive and flexible framework for data mining and knowledge discovery in human genetics.     

Genome-Wide Search for Gene-Gene Interactions in Colorectal Cancer

Genome-wide association studies (GWAS) have successfully identified a number of single-nucleotide polymorphisms (SNPs) associated with colorectal cancer (CRC) risk. However, these susceptibility loci known today explain only a small fraction of the genetic risk. Gene-gene interaction (GxG) is considered to be one source of the missing heritability. To address this, we performed a genome-wide search for pair-wise GxG associated with CRC risk using 8,380 cases and 10,558 controls in the discovery phase and 2,527 cases and 2,658 controls in the replication phase. We developed a simple, but powerful method for testing interaction, which we term the Average Risk Due to Interaction (ARDI). With this method, we conducted a genome-wide search to identify SNPs showing evidence for GxG with previously identified CRC susceptibility loci from 14 independent regions. We also conducted a genome-wide search for GxG using the marginal association screening and examining interaction among SNPs that pass the screening threshold (p<10⁻⁴). For the known locus rs10795668 (10p14), we found an interacting SNP rs367615 (5q21) with replication p = 0.01 and combined p = 4.19×10⁻⁸. Among the top marginal SNPs after LD pruning (n = 163), we identified an interaction between rs1571218 (20p12.3) and rs10879357 (12q21.1) (nominal combined p = 2.51×10⁻⁶; Bonferroni adjusted p = 0.03). Our study represents the first comprehensive search for GxG in CRC, and our results may provide new insight into the genetic etiology of CRC.     

Gene-Gene and Gene-Environment Interactions in Meta-Analysis of Genetic Association Studies

Extensive genetic studies have identified a large number of causal genetic variations in many human phenotypes; however, these could not completely explain heritability in complex diseases. Some researchers have proposed that the “missing heritability” may be attributable to gene–gene and gene–environment interactions. Because there are billions of potential interaction combinations, the statistical power of a single study is often ineffective in detecting these interactions. Meta-analysis is a common method of increasing detection power; however, accessing individual data could be difficult. This study presents a simple method that employs aggregated summary values from a “case” group to detect these specific interactions that based on rare disease and independence assumptions. However, these assumptions, particularly the rare disease assumption, may be violated in real situations; therefore, this study further investigated the robustness of our proposed method when it violates the assumptions. In conclusion, we observed that the rare disease assumption is relatively nonessential, whereas the independence assumption is an essential component. Because single nucleotide polymorphisms (SNPs) are often unrelated to environmental factors and SNPs on other chromosomes, researchers should use this method to investigate gene–gene and gene–environment interactions when they are unable to obtain detailed individual patient data.     

Anticoagulation after Anterior Myocardial Infarction and the Risk of Stroke

Background

Survivors of anterior MI are at increased risk for stroke with predilection to form ventricular thrombus. Commonly patients are discharged on dual antiplatelet therapy. Given the frequency of early coronary reperfusion and risk of bleeding, it remains uncertain whether anticoagulation offers additional utility. We examined the effectiveness of anticoagulation therapy for the prevention of stroke after anterior MI.

Methods and Findings

We performed a population-based cohort analysis of 10,383 patients who survived hospitalization for an acute MI in Ontario, Canada from April 1, 1999 to March 31, 2001. The primary outcome was four-year ischemic stroke rates compared between anterior and non-anterior MI patients. Risk factors for stroke were assessed by multivariate Cox proportional-hazards analysis. Warfarin use was determined at discharge and followed for 90 days among a subset of patients aged 66 and older (n = 1483). Among the 10,383 patients studied, 2,942 patients survived hospitalization for an anterior MI and 20% were discharged on anticoagulation therapy. Within 4 years, 169 patients (5.7%) were admitted with an ischemic stroke, half of which occurred within 1-year post-MI. There was no significant difference in stroke rate between anterior and non-anterior MI patients. The use of warfarin up to 90 days was not associated with stroke protection after anterior MI (hazard ratio [HR], 0.68; 95% confidence interval [CI], 0.37–1.26). The use of angiotensin-converting-enzyme inhibitors (HR, 0.65; 95% CI, 0.44–0.95) and beta-blockers (HR, 0.60; 95% CI, 0.41–0.87) were associated with a significant decrease in stroke risk. There was no significant difference in bleeding-related hospitalizations in patients who used warfarin for up to 90 days post-MI.

Conclusion

Many practitioners still consider a large anterior-wall MI as high risk for potential LV thrombus formation and stroke. Among a cohort of elderly patients who survived an anterior MI there was no benefit from the use of warfarin up to 90 days post-MI to prevent ischemic stroke. Our data suggests that routine anticoagulation of patients with anterior-wall MI may not be indicated. Prospective randomized trials are needed to determine the optimal antithrombin strategy for preventing this common and serious adverse outcome.     

Investigating the Role of Gene-Gene Interactions in TB Susceptibility

Tuberculosis (TB) is the second leading cause of mortality from infectious disease worldwide. One of the factors involved in developing disease is the genetics of the host, yet the field of TB susceptibility genetics has not yielded the answers that were expected. A commonly posited explanation for the missing heritability of complex disease is gene-gene interactions, also referred to as epistasis. In this study we investigate the role of gene-gene interactions in genetic susceptibility to TB using a cohort recruited from a high TB incidence community from Cape Town, South Africa. Our discovery data set incorporates genotypes from a large a number of candidate gene studies as well as genome-wide data. After limiting our search space to pairs of putative TB susceptibility genes, as well as pairs of genes that have been curated in online databases as potential interactors, we use statistical modelling to identify pairs of interacting SNPs. We attempt to validate the top models identified in our discovery data set using an independent genome-wide TB case-control data set from The Gambia. A number of models were successfully validated, indicating that interplay between the NRG1 - NRG3, GRIK1 - GRIK3 and IL23R - ATG4C gene pairs may modify susceptibility to TB. Gene pairs involved in the NF-κB pathway were also identified in the discovery data set (SFTPD - NOD2, ISG15 - TLR8 and NLRC5 - IL12RB1), but could not be tested in the Gambian study group due to lack of overlapping data.     

Gene-Gene Interactions Lead to Higher Risk for Development of Type 2 Diabetes in an Ashkenazi Jewish Population

Background

Evidence has accumulated that multiple genetic and environmental factors play important roles in determining susceptibility to type 2 diabetes (T2D). Although variants from candidate genes have become prime targets for genetic analysis, few studies have considered their interplay. Our goal was to evaluate interactions among SNPs within genes frequently identified as associated with T2D.

Methods/Principal Findings

Logistic regression was used to study interactions among 4 SNPs, one each from HNF4A[rs1884613], TCF7L2[rs12255372], WFS1[rs10010131], and KCNJ11[rs5219] in a case-control Ashkenazi sample of 974 diabetic subjects and 896 controls. Nonparametric multifactor dimensionality reduction (MDR) and generalized MDR (GMDR) were used to confirm findings from the logistic regression analysis. HNF4A and WFS1 SNPs were associated with T2D in logistic regression analyses [P<0.0001, P<0.0002, respectively]. Interaction between these SNPs were also strong using parametric or nonparametric methods: the unadjusted odds of being affected with T2D was 3 times greater in subjects with the HNF4A and WFS1 risk alleles than those without either (95% CI = [1.7–5.3]; P≤0.0001). Although the univariate association between the TCF7L2 SNP and T2D was relatively modest [P = 0.02], when paired with the HNF4A SNP, the OR for subjects with risk alleles in both SNPs was 2.4 [95% CI = 1.7–3.4; P≤0.0001]. The KCNJ11 variant reached significance only when paired with either the HNF4A or WFSI SNPs: unadjusted ORs were 2.0 [95% CI = 1.4–2.8; P≤0.0001] and 2.3 [95% CI = 1.2-4.4; P≤0.0001], respectively. MDR and GMDR results were consistent with the parametric findings.

Conclusions

These results provide evidence of strong independent associations between T2D and SNPs in HNF4A and WFS1 and their interaction in our Ashkenazi sample. We also observed an interaction in the nonparametric analysis between the HNF4A and KCNJ11 SNPs (P≤0.001), demonstrating that an independently non-significant variant may interact with another variant resulting in an increased disease risk.     

Polymorphism of the Apolipoprotein E Gene and Risk of Myocardial Infarction

The apolipoprotein E (ApoE) gene polymorphism resulting from nucleotide substitutions in exon 4 was analyzed in Russian and Tatar patients with myocardial infarction (MI) from Bashkortostan. Alleles 2, 3, and 4 were identified by PCR. The genotype frequency distribution proved to be age-dependent in healthy Russians, genotype 2/3 increasing in frequency in subjects over 45. Russians who suffered MI under 45 had lower frequencies of genotype E3/3 (50.00% vs. 75.47% in controls of the same age, = 0.013, OR = 0.33) and allele 3 (72.12% vs. 85.85%, = 0.020, OR = 0.43) and a higher frequency of allele 4 (22.12% vs. 10.38%, = 0.030, OR = 2.45). Russians who suffered MI complicated by cardiogenic shock (CS) had a significantly higher frequency of genotype 3/4 and lower frequencies of genotype 3/3 and allele 3 as compared with MI patients without CS. In Tatars, genotype 4/4 occurred at a frequency of 14.29% in patients who suffered MI under 45, and was not detected in healthy subjects of the same age ( = 0.024, OR = 17.85). Thus, the ApoE polymorphism was associated with higher risk of MI in Russians and Tatars under 45.     

Genetic Variants from Lipid-Related Pathways and Risk for Incident Myocardial Infarction

Background

Circulating lipids levels, as well as several familial lipid metabolism disorders, are strongly associated with initiation and progression of atherosclerosis and incidence of myocardial infarction (MI).

Objectives

We hypothesized that genetic variants associated with circulating lipid levels would also be associated with MI incidence, and have tested this in three independent samples.

Setting and Subjects

Using age- and sex-adjusted additive genetic models, we analyzed 554 single nucleotide polymorphisms (SNPs) in 41 candidate gene regions proposed to be involved in lipid-related pathways potentially predisposing to incidence of MI in 2,602 participants of the Swedish Twin Register (STR; 57% women). All associations with nominal P<0.01 were further investigated in the Uppsala Longitudinal Study of Adult Men (ULSAM; N = 1,142).

Results

In the present study, we report associations of lipid-related SNPs with incident MI in two community-based longitudinal studies with in silico replication in a meta-analysis of genome-wide association studies. Overall, there were 9 SNPs in STR with nominal P-value <0.01 that were successfully genotyped in ULSAM. rs4149313 located in ABCA1 was associated with MI incidence in both longitudinal study samples with nominal significance (hazard ratio, 1.36 and 1.40; P-value, 0.004 and 0.015 in STR and ULSAM, respectively). In silico replication supported the association of rs4149313 with coronary artery disease in an independent meta-analysis including 173,975 individuals of European descent from the CARDIoGRAMplusC4D consortium (odds ratio, 1.03; P-value, 0.048).

Conclusions

rs4149313 is one of the few amino acid changing variants in ABCA1 known to associate with reduced cholesterol efflux. Our results are suggestive of a weak association between this variant and the development of atherosclerosis and MI.     

老年急性心肌梗死患者并发肺部感染的危险因素分析及护理对策

目的探讨老年急性心肌梗死(AMI)患者并发肺部感染的危险因素及护理对策。方法回顾性分析我院心内科2012年1月~2015年1月收治的老年AMI患者160例的临床资料,根据患者有无并发肺部感染分为感染组(观察组,n=36)和非感染组(对照组,n=124),比较两组患者的性别、年龄、吸烟史、左室射血分数(LVEF)、基础疾病等变量,研究上述变量与肺部感染的相关性,进一步分析AMI患者并发肺部感染的独立危险因素,比较两组患者住院时间和死亡率。结果观察组在年龄、合并糖尿病、吸烟等变量较对照组明显升高,LVEF值较对照组明显下降,差异均有统计学意义(P0.05)。将这些变量纳入多元Logistic回归分析显示,年龄、低EF水平及合并糖尿病是老年AMI患者合并肺部感染的独立危险因素。观察组住院时间为(24.5±2.6)天明显长于对照组的(18.2±3.8)天,病死率为25%明显高于对照组的15.2%,差异均有统计学意义(均P0.05)。结论高龄、低LVEF、合并糖尿病是老年AMI患者并发肺部感染的独立危险因素,对合并上述危险因素患者应实施针对性的护理措施,从而降低老年心梗的病死率。     

Analysis of Gene-Gene Interactions among Common Variants in Candidate Cardiovascular Genes in Coronary Artery Disease

ObjectiveOnly a small fraction of coronary artery disease (CAD) heritability has been explained by common variants identified to date. Interactions between genes of importance to cardiovascular regulation may account for some of the missing heritability of CAD. This study aimed to investigate the role of gene-gene interactions in common variants in candidate cardiovascular genes in CAD.ConclusionsModerately large additive interactions between common SNPs in genes relevant to cardiovascular disease do not appear to play a major role in genetic predisposition to CAD. The role of genetic interactions amongst less common SNPs and with medium and small magnitude effects remain to be investigated.     

A Simple and Computationally Efficient Approach to Multifactor Dimensionality Reduction Analysis of Gene-Gene Interactions for Quantitative Traits

We present an extension of the two-class multifactor dimensionality reduction (MDR) algorithm that enables detection and characterization of epistatic SNP-SNP interactions in the context of a quantitative trait. The proposed Quantitative MDR (QMDR) method handles continuous data by modifying MDR’s constructive induction algorithm to use a T-test. QMDR replaces the balanced accuracy metric with a T-test statistic as the score to determine the best interaction model. We used a simulation to identify the empirical distribution of QMDR’s testing score. We then applied QMDR to genetic data from the ongoing prospective Prevention of Renal and Vascular End-Stage Disease (PREVEND) study.     

主动脉内球囊反搏在急性心肌梗死中的治疗作用研究

目的:评价主动脉内气囊反搏术对高危心梗患者住院期间的治疗作用及长期预后的影响.方法:回顾性分析我院2006年到2010年间50例心梗病人,其中随机选取25例应用IABP,其余25例为对照组,比较2组患者在入院时和5天后血压、心肌损伤标志物、BNP的差异,并随访6个月时左室射血功能(EF).结果:IABP组在5天后血压明显高于对照组,心肌损伤标志物和BNP恢复较对照组增快,6月时EF较对照组提高明显,均具有统计学意义.结论:在高危心梗患者行PCI介入治疗时,尽早应用IABP能改善患者心功能.     

Allelic Combinations of Immune Response Genes and Risk of Development of Myocardial Infarction

Myocardial infarction (MI) is a multifactorial polygenic disease. It develops because of the complex interaction between many environmental and genetic factors. In this paper, we have studied associations of MI and allele combinations of 17 polymorphic markers of immune response genes in an ethnically homogeneous group of Tatars. The material for analysis was DNA samples of patients (286 men) with onset of MI at the age of 30 to 60 years and 301 essentially healthy men of the control group. Using the APSampler algorithm, we obtained allele combinations with the increased risk of MI in which allele variants CX3CR1*M (rs3732378), VCAM1*C (rs3917010), ICAM1*E (rs5498), LTA*A (rs909253), and TNFRSF1B*M (rs1061622) occurred the most often.     

Combined Biomarker Analysis for Risk of Acute Kidney Injury in Patients with ST-Segment Elevation Myocardial Infarction

Background

Acute kidney injury (AKI) complicating ST-segment elevation myocardial infarction (STEMI) increases subsequent morbidity and mortality. We combined the biomarkers of heart failure (HF; B-type natriuretic peptide [BNP] and soluble ST2 [sST2]) and renal injury (NGAL [neutrophil gelatinase-associated lipocalin] and cystatin C) in predicting the development of AKI in patients with STEMI undergoing primary percutaneous coronary intervention (PCI).

Methods and Results

From March 2010 to September 2013, 189 STEMI patients were sequentially enrolled and serum samples were collected at presentation for BNP, sST2, NGAL and cystatin C analysis. 37 patients (19.6%) developed AKI of varying severity within 48 hours of presentation. Univariate analysis showed age, Killip class ≥2, hypertension, white blood cell counts, hemoglobin, estimated glomerular filtration rate, blood urea nitrogen, creatinine, and all the four biomarkers were predictive of AKI. Serum levels of the biomarkers were correlated with risk of AKI and the Acute Kidney Injury Network (AKIN) stage and all significantly discriminated AKI (area under the receiver operating characteristic [ROC] curve: BNP: 0.86, sST2: 0.74, NGAL: 0.75, cystatin C: 0.73; all P < 0.05). Elevation of ≥2 of the biomarkers higher than the cutoff values derived from the ROC analysis improved AKI risk stratification, regardless of the creatine level (creatinine < 1.24 mg/dL: odds ratio [OR] 11.25, 95% confidence interval [CI] 1.63-77.92, P = 0.014; creatinine ≥ 1.24: OR 15.0, 95% CI 1.23-183.6, P = 0.034).

Conclusions

In this study of STEMI patients undergoing primary PCI, the biomarkers of heart failure (BNP and sST2) and renal injury (NGAL and cystatin C) at presentation were predictive of AKI. High serum levels of the biomarkers were associated with an elevated risk and more advanced stage of AKI. Regardless of the creatinine level, elevation of ≥2 of the biomarkers higher than the cutoff values indicated a further rise in AKI risk. Combined biomarker approach may assist in risk stratification of AKI in patients with STEMI.