Prediction of Quantitative Traits Using Common Genetic Variants: Application to Body Mass Index

Investigation of prediction accuracy and the impact of sample size, ancestry, and tissue in transcriptome-wide association studies

In transcriptome-wide association studies (TWAS), gene expression values are predicted using genotype data and tested for association with a phenotype. The power of this approach to detect associations relies, at least in part, on the accuracy of the prediction. Here we compare the prediction accuracy of six different methods-LASSO, Ridge regression, Elastic net, Best Linear Unbiased Predictor, Bayesian Sparse Linear Mixed Model, and Random Forests-by performing cross-validation using data from the Geuvadis Project. We also examine prediction accuracy (a) at different sample sizes, (b) when ancestry of the prediction model training and testing populations is different, and (c) when the tissue used to train the model is different from the tissue to be predicted. We find that, for most genes, the expression cannot be accurately predicted, but in general sparse statistical models tend to outperform polygenic models at prediction. Average prediction accuracy is reduced when the model training set size is reduced or when predicting across ancestries and is marginally reduced when predicting across tissues. We conclude that using sparse statistical models and the development of large reference panels across multiple ethnicities and tissues will lead to better prediction of gene expression, and thus may improve TWAS power.

Signatures of selection analysis using whole-genome sequence data reveals novel candidate genes for pony and light horse types

Natural selection and domestication have shaped modern horse populations, resulting in a vast range of phenotypically diverse breeds. Horse breeds are classified into three types (pony, light, and draft) generally based on their body type. Understanding the genetic basis of horse type variation and selective pressures related to the evolutionary trend can be particularly important for current selection strategies. Whole-genome sequences were generated for 14 pony and 32 light horses to investigate the genetic signatures of selection of the horse type in pony and light horses. In the overlapping extremes of the fixation index and nucleotide diversity results, we found novel genomic signatures of selective sweeps near key genes previously implicated in body measurements including C4ORF33, CRB1, CPN1, FAM13A, and FGF12 that may influence variation in pony and light horse types. This study contributes to a better understanding of the genetic background of differences between pony and light horse types.

Weighting approaches for a genetic risk score and an oxidative stress score for predicting the incidence of obesity

BACKGROUND

We aimed to predict the incidence of obesity in a Korean population using a genetic risk score (GRS) constructed with obesity-related single nucleotide polymorphisms (SNPs) along with an oxidative stress score (OSS).

METHODS

A total of 9460 Korean subjects and 356 974 SNPs were included. The GRS was constructed using three significant obesity-related SNP loci, and the OSS was calculated with three reliable oxidative stress biomarkers.

RESULTS

The GRS showed a more significant association with increased obesity (OR = 2.879) than did individual SNPs after adjusting for age and sex. Three oxidative stress biomarkers, including malondialdehyde, oxidized low-density lipoprotein, and 8-epi-prostaglandin F_2α , showed significantly high levels in the obese group. The OSS, which was the sum of each oxidative stress biomarker score, showed a markedly high association with the incidence of obesity, with an OR of 3.213. Based on the results of the regression tests and a receiver-operating characteristic (ROC) curve analysis, we found that HOMA-IR, high-sensitivity C-reactive protein (hs-CRP), the GRS, and the OSS were the most relevant factors for the increased risk of obesity and were significantly associated with the incidence of obesity. The area under the ROC curve was improved when the GRS was added to the model (from 74.2% to 75.1%).

CONCLUSIONS

We first identified that subjects with an obesity GRS and a high OSS might have a higher risk of obesity. Our findings and weighting approaches were effective in predicting the incidence of obesity; furthermore, the GRS is a relevant factor that significantly predicts the risk of obesity.

Pure additive contribution of genetic variants to a risk prediction model using propensity score matching: application to type 2 diabetes

The achievements of genome-wide association studies have suggested ways to predict diseases, such as type 2 diabetes (T2D), using single-nucleotide polymorphisms (SNPs). Most T2D risk prediction models have used SNPs in combination with demographic variables. However, it is difficult to evaluate the pure additive contribution of genetic variants to classically used demographic models. Since prediction models include some heritable traits, such as body mass index, the contribution of SNPs using unmatched case-control samples may be underestimated. In this article, we propose a method that uses propensity score matching to avoid underestimation by matching case and control samples, thereby determining the pure additive contribution of SNPs. To illustrate the proposed propensity score matching method, we used SNP data from the Korea Association Resources project and reported SNPs from the genome-wide association study catalog. We selected various SNP sets via stepwise logistic regression (SLR), least absolute shrinkage and selection operator (LASSO), and the elastic-net (EN) algorithm. Using these SNP sets, we made predictions using SLR, LASSO, and EN as logistic regression modeling techniques. The accuracy of the predictions was compared in terms of area under the receiver operating characteristic curve (AUC). The contribution of SNPs to T2D was evaluated by the difference in the AUC between models using only demographic variables and models that included the SNPs. The largest difference among our models showed that the AUC of the model using genetic variants with demographic variables could be 0.107 higher than that of the corresponding model using only demographic variables.

BMI prediction within a Korean population

BACKGROUND

Body Mass Index (BMI) is widely regarded as an important clinical trait for obesity and other diseases such as Type 2 diabetes, coronary heart disease, and osteoarthritis.

METHODS

This study uses 6,011 samples of genotype data from ethnic Korean subjects. The data was retrieved from the Korea Association Resource. To identify the BMI-related markers within the Korean population, we collected genome-wide association study (GWAS) markers using a GWAS catalog and also obtained other markers from nearby regions. Of the total 6,011 samples, 5,410 subjects were used as part of a single nucleotide polymorphism (SNP) selection set in order to identify the overlapping BMI-associated SNPs within a 10-fold cross validation.

RESULTS

We selected nine SNPs (rs12566985 (FPGT-TNNI3K), rs6545809 (ADCY3), rs2943634 (located near LOC646736), rs734597 (located near TFAP2B), rs11030104 (BDNF), rs7988412 (GTF3A), rs2241423 (MAP2K5), rs7202116 (FTO), and rs6567160 (located near LOC105372152) to assist in BMI prediction. The calculated weighted genetic risk scores based on the selected 9 SNPs within the SNP selection set were applied to the final validation set consisting of 601 samples. Our results showed upward trends in the BMI values (P < 0.0001) within the 10-fold cross validation process for R² > 0.22. These trends were also observed within the validation set for all subjects, as well as within the validation sets divided by gender (P < 0.0001, R² > 0.46).

DISCUSSION

The set of nine SNPs identified in this study may be useful for prospective predictions of BMI.