Cross-trait prediction accuracy of summary statistics in genome-wide association studies

In the era of big data, univariate models have widely been used as a workhorse tool for quickly producing marginal estimators; and this is true even when in a high-dimensional dense setting, in which many features are “true,” but weak signals. Genome-wide association studies (GWAS) epitomize this type of setting. Although the GWAS marginal estimator is popular, it has long been criticized for ignoring the correlation structure of genetic variants (i.e., the linkage disequilibrium [LD] pattern). In this paper, we study the effects of LD pattern on the GWAS marginal estimator and investigate whether or not additionally accounting for the LD can improve the prediction accuracy of complex traits. We consider a general high-dimensional dense setting for GWAS and study a class of ridge-type estimators, including the popular marginal estimator and the best linear unbiased prediction (BLUP) estimator as two special cases. We show that the performance of GWAS marginal estimator depends on the LD pattern through the first three moments of its eigenvalue distribution. Furthermore, we uncover that the relative performance of GWAS marginal and BLUP estimators highly depends on the ratio of GWAS sample size over the number of genetic variants. Particularly, our finding reveals that the marginal estimator can easily become near-optimal within this class when the sample size is relatively small, even though it ignores the LD pattern. On the other hand, BLUP estimator has substantially better performance than the marginal estimator as the sample size increases toward the number of genetic variants, which is typically in millions. Therefore, adjusting for the LD (such as in the BLUP) is most needed when GWAS sample size is large. We illustrate the importance of our results by using the simulated data and real GWAS.     

Genomic architecture of autism from comprehensive whole-genome sequence annotation

1. Download : Download high-res image (172KB)
2. Download : Download full-size image

     

Using the tools of genetic epidemiology to understand sex differences in neuropsychiatric disorders

Many neuropsychiatric disorders exhibit differences in prevalence, age of onset, symptoms or course of illness between males and females. For the most part, the origins of these differences are not well understood. In this article, we provide an overview of sex differences in psychiatric disorders including autism spectrum disorder (ASD), attention deficit/hyperactivity disorder (ADHD), anxiety, depression, alcohol and substance abuse, schizophrenia, eating disorders and risk of suicide. We discuss both genetic and nongenetic mechanisms that have been hypothesized to underlie these differences, including ascertainment bias, environmental stressors, X‐ or Y‐linked risk loci, and differential liability thresholds in males and females. We then review the use of twin, family and genome‐wide association approaches to study potential genetic mechanisms of sex differences and the extent to which these designs have been employed in studies of psychiatric disorders. We describe the utility of genetic epidemiologic study designs, including classical twin and family studies, large‐scale studies of population registries, derived recurrence risks, and molecular genetic analyses of genome‐wide variation that may enhance our understanding sex differences in neuropsychiatric disorders.     

Ruminal fermentation characteristics and related feeding values of compound feeds and their constituting single feeds studied by using in vitrotechniques

Single concentrate feeds are mixed together forming compound feeds for cattle. However, knowledge regarding the potential interactions (associative effects) between the feeding values of single feeds in compound feeds is lacking. The main objective of the present study was to evaluate ruminal fermentation characteristics and feeding values of eight industrially produced compound feeds in mash form from their constituent single feeds for dairy cows through in vitroassays. Additivity was given for gas production (GP), digestibility of organic matter (dOM) and utilisable CP at the duodenum (uCP). Additivity of CP fractions (determined using the Cornell Net Carbohydrate and Protein System (CNCPS)) was dependent on the fraction and compound feed type; however, the effective degradation calculated from CP fractions (ED_CNCPS) showed additivity. Additivity was not given for intestinal digestibility of rumen-undegraded protein (ID_RUP) for five out of eight compound feeds. Precise calculation of metabolisable energy (ME) of compound feeds from ME of single feeds was possible when using the same ME equations for all single and compound feeds. Compound feeds are often provided in pellet form; therefore, our second objective was to evaluate the effects of pelleting on ruminal fermentation characteristics and feeding values of compound feeds. Pelleting affected GP at 24 h (GP₂₄; up to 2.4 ml/200 mg DM), dOM (up to 2.3 percentage point (pp)) and ME (up to 0.3 MJ/kg DM), but these differences were overall small. More considerable effects of pelleting were observed for uCP, which was increased in all compound feeds except the two with the highest CP concentrations. The ID_RUPwas lower in most compound feeds following pelleting (up to 15 pp). Pelleting also affected CP fractions in a non-systematic way. Overall, the effects of pelleting were not considerable, which could be because pelleting conditions were mild. Our third objective was to compare in situruminal CP degradation (ED_{IN_SITU}) of compound feeds with ED using two prediction methods based on CP fractions. ED_{IN_SITU}reference data were obtained from a companion study using the same feeds. Prediction accuracy of ED_{IN_SITU}and ED_CNCPSwas variable and depended on the compound feed and prediction method. However, future studies are needed as to date not enough data are published to draw overall conclusions for the prediction of ED_{IN_SITU}from CP fractions.     

基于不同可加性方法的黑龙江省红松人工林林分生物量模型

大尺度估算森林生物量一直是人们关注的焦点,而构建林分水平的生物量模型是一种估算森林乔木层生物量的方法。本研究基于聚合法1、聚合法2、平差法、分解法构建红松人工林林分生物量模型,并对比分析4种可加性方法的预测精度,为黑龙江省红松人工林的生物量预测提供科学依据。各模型均使用权函数来消除各模型的异方差,并以留一交叉验证法(LOOCV)作为各模型的检验方法。结果表明: 平差法的整体预测能力略优于聚合法1、聚合法2和分解法,预测精度排序为平差法>聚合法1>聚合法2>分解法;分别对比不同林分断面积的预测能力时,4种可加性方法的预测精度不一致。当红松人工林的林分断面积分布于0～10或50～60 m²·hm^-2区间时,建议采用分解法的参数估计值,而林分断面积分布于其他区间时,建议采用平差法的参数估计值。