An Effective Method to Identify Heritable Components from Multivariate Phenotypes

Heritability estimation for a linear combination of phenotypes via ridge regression

MOTIVATION

The joint analysis of multiple phenotypes is important in many biological studies, such as plant and animal breeding. The heritability estimation for a linear combination of phenotypes is designed to account for correlation information. Existing methods for estimating heritability mainly focus on single phenotypes under random-effect models. These methods also require some stringent conditions, which calls for a more flexible and interpretable method for estimating heritability. Fixed-effect models emerge as a useful alternative.

RESULTS

In this paper, we propose a novel heritability estimator based on multivariate ridge regression for linear combinations of phenotypes, yielding accurate estimates in both sparse and dense cases. Under mild conditions in the high-dimensional setting, the proposed estimator appears to be consistent and asymptotically normally distributed. Simulation studies show that the proposed estimator is promising under different scenarios. Compared with independently combined heritability estimates in the case of multiple phenotypes, the proposed method significantly improves the performance by considering correlations among those phenotypes. We further demonstrate its application in heritability estimation and correlation analysis for the Oryza sativa rice dataset.

AVAILABILITY AND IMPLEMENTATION

An R package implementing the proposed method is available at https://github.com/xg-SUFE1/MultiRidgeVar, where covariance estimates are also given together with heritability estimates.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Where the genome meets the connectome: Understanding how genes shape human brain connectivity

The integration of modern neuroimaging methods with genetically informative designs and data can shed light on the molecular mechanisms underlying the structural and functional organization of the human connectome. Here, we review studies that have investigated the genetic basis of human brain network structure and function through three complementary frameworks: (1) the quantification of phenotypic heritability through classical twin designs; (2) the identification of specific DNA variants linked to phenotypic variation through association and related studies; and (3) the analysis of correlations between spatial variations in imaging phenotypes and gene expression profiles through the integration of neuroimaging and transcriptional atlas data. We consider the basic foundations, strengths, limitations, and discoveries associated with each approach. We present converging evidence to indicate that anatomical connectivity is under stronger genetic influence than functional connectivity and that genetic influences are not uniformly distributed throughout the brain, with phenotypic variation in certain regions and connections being under stronger genetic control than others. We also consider how the combination of imaging and genetics can be used to understand the ways in which genes may drive brain dysfunction in different clinical disorders.

VIGAN: Missing View Imputation with Generative Adversarial Networks

In an era when big data are becoming the norm, there is less concern with the quantity but more with the quality and completeness of the data. In many disciplines, data are collected from heterogeneous sources, resulting in multi-view or multi-modal datasets. The missing data problem has been challenging to address in multi-view data analysis. Especially, when certain samples miss an entire view of data, it creates the missing view problem. Classic multiple imputations or matrix completion methods are hardly effective here when no information can be based on in the specific view to impute data for such samples. The commonly-used simple method of removing samples with a missing view can dramatically reduce sample size, thus diminishing the statistical power of a subsequent analysis. In this paper, we propose a novel approach for view imputation via generative adversarial networks (GANs), which we name by VIGAN. This approach first treats each view as a separate domain and identifies domain-to-domain mappings via a GAN using randomly-sampled data from each view, and then employs a multi-modal denoising autoencoder (DAE) to reconstruct the missing view from the GAN outputs based on paired data across the views. Then, by optimizing the GAN and DAE jointly, our model enables the knowledge integration for domain mappings and view correspondences to effectively recover the missing view. Empirical results on benchmark datasets validate the VIGAN approach by comparing against the state of the art. The evaluation of VIGAN in a genetic study of substance use disorders further proves the effectiveness and usability of this approach in life science.

Collaborative Phenotype Inference from Comorbid Substance Use Disorders and Genotypes

Data in large-scale genetic studies of complex human diseases, such as substance use disorders, are often incomplete. Despite great progress in genotype imputation, e.g., the IMPUTE2 method, considerably less progress has been made in inferring phenotypes. We designed a novel approach to integrate individuals' comorbid conditions with their genotype data to infer missing (unreported) diagnostic criteria of a disorder. The premise of our approach derives from correlations among symptoms and the shared biological bases of concurrent disorders such as co-dependence on cocaine and opioids. We describe a matrix completion method to construct a bi-linear model based on the interactions of genotypes and known symptoms of related disorders to infer unknown values of another set of symptoms or phenotypes. An efficient stochastic and parallel algorithm based on the linearized alternating direction method of multipliers was developed to solve the proposed optimization problem. Empirical evaluation of the approach in comparison with other advanced data matrix completion methods via a case study shows that it both significantly improves imputation accuracy and provides greater computational efficiency.

Pharmacotherapy for smoking cessation: effects by subgroup defined by genetically informed biomarkers

BACKGROUND

Smoking cessation therapies are not effective for all smokers, and researchers are interested in identifying those subgroups of individuals (e.g. based on genotype) who respond best to specific treatments.

OBJECTIVES

To assess whether quit rates vary by genetically informed biomarkers within pharmacotherapy treatment arms and as compared with placebo. To assess the effects of pharmacotherapies for smoking cessation in subgroups of smokers defined by genotype for identified genome-wide significant polymorphisms.

SEARCH METHODS

We searched the Cochrane Tobacco Addiction Group specialised register, clinical trial registries, and genetics databases for trials of pharmacotherapies for smoking cessation from inception until 16 August 2016.

SELECTION CRITERIA

We included randomised controlled trials (RCTs) that recruited adult smokers and reported pharmacogenomic analyses from trials of smoking cessation pharmacotherapies versus controls. Eligible trials included those with data on a priori genome-wide significant (P < 5 × 10-8) single-nucleotide polymorphisms (SNPs), replicated non-SNPs, and/or the nicotine metabolite ratio (NMR), hereafter collectively described as biomarkers.

DATA COLLECTION AND ANALYSIS

We used standard methodological procedures expected by Cochrane. The primary outcome was smoking abstinence at six months after treatment. The secondary outcome was abstinence at end of treatment (EOT). We conducted two types of meta-analyses- one in which we assessed smoking cessation of active treatment versus placebo within genotype groups, and another in which we compared smoking cessation across genotype groups within treatment arms. We carried out analyses separately in non-Hispanic whites (NHWs) and non-Hispanic blacks (NHBs). We assessed heterogeneity between genotype groups using T², I², and Cochrane Q statistics.

MAIN RESULTS

Analyses included 18 trials including 9017 participants, of whom 6924 were NHW and 2093 NHB participants. Data were available for the following biomarkers: nine SNPs (rs1051730 (CHRNA3); rs16969968, rs588765, and rs2036527 (CHRNA5); rs3733829 and rs7937 (in EGLN2, near CYP2A6); rs1329650 and rs1028936 (LOC100188947); and rs215605 (PDE1C)), two variable number tandem repeats (VNTRs; DRD4 and SLC6A4), and the NMR. Included data produced a total of 40 active versus placebo comparisons, 16 active versus active comparisons, and 64 between-genotype comparisons within treatment arms.For those meta-analyses showing statistically significant heterogeneity between genotype groups, we found the quality of evidence (GRADE) to be generally moderate. We downgraded quality most often because of imprecision or risk of bias due to potential selection bias in genotyping trial participants. Comparisons of relative treatment effects by genotypeFor six-month abstinence, we found statistically significant heterogeneity between genotypes (rs16969968) for nicotine replacement therapy (NRT) versus placebo at six months for NHB participants (P = 0.03; n = 2 trials), but not for other biomarkers or treatment comparisons. Six-month abstinence was increased in the active NRT group as compared to placebo among participants with a GG genotype (risk ratio (RR) 1.47, 95% confidence interval (CI) 1.07 to 2.03), but not in the combined group of participants with a GA or AA genotype (RR 0.43, 95% CI 0.15 to 1.26; ratio of risk ratios (RRR) GG vs GA or AA of 3.51, 95% CI 1.19 to 10.3). Comparisons of treatment effects between genotype groups within pharmacotherapy randomisation armsFor those receiving active NRT, treatment was more effective in achieving six-month abstinence among individuals with a slow NMR than among those with a normal NMR among NHW and NHB combined participants (normal NMR vs slow NMR: RR 0.54, 95% CI 0.37 to 0.78; n = 2 trials). We found no such differences in treatment effects between genotypes at six months for any of the other biomarkers among individuals who received pharmacotherapy or placebo.

AUTHORS' CONCLUSIONS

We did not identify widespread differential treatment effects of pharmacotherapy based on genotype. Some genotype groups within certain ethnic groups may benefit more from NRT or may benefit less from the combination of bupropion with NRT. The reader should interpret these results with caution because none of the statistically significant meta-analyses included more than two trials per genotype comparison, many confidence intervals were wide, and the quality of this evidence (GRADE) was generally moderate. Although we found evidence of superior NRT efficacy for NMR slow versus normal metabolisers, because of the lack of heterogeneity between NMR groups, we cannot conclude that NRT is more effective for slow metabolisers. Access to additional data from multiple trials is needed, particularly for comparisons of different pharmacotherapies.