Improving the Power of Sib Pair Quantitative Trait Loci Detection by Phenotype Winsorization

Fernández, José R.; Etzel, Carol J.; Beasley, T. Mark; Shete, Sanjay; Amos, Christopher I.; Allison, David B.

doi:10.1159/000057984

Cited by 33 publications

(27 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For specific values of c S and c D , NHE reduces to OHE and RHE [1]. Variations of this NHE have been independently published by several groups [8][9][10][11][12].…”

Section: Introductionmentioning

confidence: 94%

Effect of Box-Cox Transformation on Power of Haseman-Elston and Maximum-Likelihood Variance Components Tests to Detect Quantitative Trait Loci

Etzel¹,

Shete²,

Beasley³

et al. 2003

Hum Hered

View full text Add to dashboard Cite

Non-normality of the phenotypic distribution can affect power to detect quantitative trait loci in sib pair studies. Previously, we observed that Winsorizing the sib pair phenotypes increased the power of quantitative trait locus (QTL) detection for both Haseman-Elston (HE) least-squares tests [Hum Hered 2002;53:59–67] and maximum likelihood-based variance components (MLVC) analysis [Behav Genet (in press)]. Winsorizing the phenotypes led to a slight increase in type 1 error in H-E tests and a slight decrease in type I error for MLVC analysis. Herein, we considered transforming the sib pair phenotypes using the Box-Cox family of transformations. Data were simulated for normal and non-normal (skewed and kurtic) distributions. Phenotypic values were replaced by Box-Cox transformed values. Twenty thousand replications were performed for three H-E tests of linkage and the likelihood ratio test (LRT), the Wald test and other robust versions based on the MLVC method. We calculated the relative nominal inflation rate as the ratio of observed empirical type 1 error divided by the set α level (5, 1 and 0.1% α levels). MLVC tests applied to non-normal data had inflated type I errors (rate ratio greater than 1.0), which were controlled best by Box-Cox transformation and to a lesser degree by Winsorizing. For example, for non-transformed, skewed phenotypes (derived from a χ² distribution with 2 degrees of freedom), the rates of empirical type 1 error with respect to set α level = 0.01 were 0.80, 4.35 and 7.33 for the original H-E test, LRT and Wald test, respectively. For the same α level = 0.01, these rates were 1.12, 3.095 and 4.088 after Winsorizing and 0.723, 1.195 and 1.905 after Box-Cox transformation. Winsorizing reduced inflated error rates for the leptokurtic distribution (derived from a Laplace distribution with mean 0 and variance 8). Further, power (adjusted for empirical type 1 error) at the 0.01 α level ranged from 4.7 to 17.3% across all tests using the non-transformed, skewed phenotypes, from 7.5 to 20.1% after Winsorizing and from 12.6 to 33.2% after Box-Cox transformation. Likewise, power (adjusted for empirical type 1 error) using leptokurtic phenotypes at the 0.01 α level ranged from 4.4 to 12.5% across all tests with no transformation, from 7 to 19.2% after Winsorizing and from 4.5 to 13.8% after Box-Cox transformation. Thus the Box-Cox transformation apparently provided the best type 1 error control and maximal power among the procedures we considered for analyzing a non-normal, skewed distribution (χ²) while Winzorizing worked best for the non-normal, kurtic distribution (Laplace). We repeated the same simulations using a larger sample size (200 sib pairs) and found similar results.

show abstract

“…For specific values of c S and c D , NHE reduces to OHE and RHE [1]. Variations of this NHE have been independently published by several groups [8][9][10][11][12].…”

Section: Introductionmentioning

confidence: 94%

Effect of Box-Cox Transformation on Power of Haseman-Elston and Maximum-Likelihood Variance Components Tests to Detect Quantitative Trait Loci

Etzel¹,

Shete²,

Beasley³

et al. 2003

Hum Hered

View full text Add to dashboard Cite

show abstract

“…For the TG trait, we found some very extreme outliers. Therefore, we used winsorization to bring the upper and lower 5% of the TG values closer to the trait mean (24,25). As described by Shete et al (25), winsorization increases the power to detect linkage and reduces the bias in estimation of the major VC.…”

Section: Error Checking and Data Handlingmentioning

confidence: 99%

A genome-wide linkage scan identifies multiple chromosomal regions influencing serum lipid levels in the population on the Samoan islands

Åberg

Dai

Sun

et al. 2008

Journal of Lipid Research

View full text Add to dashboard Cite

show abstract

“…Residuals included adjustment for age, body mass index, HDL cholesterol, systolic BP, hypertension treatment, diabetes, and current smoking. A residual value Ͼ2.5 SD from the mean was reduced to the 2.5 SD value to improve the skewness and kurtosis of the data (21). In addition, all GFR and CRCL values Ͼ200 were considered to be 200.…”

Section: Statistical Analysesmentioning

confidence: 99%

Genomewide Linkage Analysis to Serum Creatinine, GFR, and Creatinine Clearance in a Community-Based Population

Fox

Yang²,

Cupples³

et al. 2004

Journal of the American Society of Nephrology

163

118

View full text Add to dashboard Cite

Abstract. Kidney disease is a risk factor for the development of cardiovascular disease, all-cause mortality, and ESRD. It is not known to what extent genetic factors play a role in the development of kidney disease in the general population. Multipoint variance components linkage analysis was performed using Genehunter on 330 families from the Framingham Heart Study offspring cohort, using a 10-cM genomewide scan for serum creatinine, GFR, and creatinine clearance (CRCL) measured from 1998 to 2001. GFR was estimated using the simplified Modification of Diet in Renal Disease Study equation, and CRCL was estimated using the Cockcroft-Gault equation. Covariates in the adjustment included age, gender, body mass index, diabetes, systolic BP, hypertension treatment, tobacco use, and HDL cholesterol. Overall, 1224 subjects (52% women), mean age 59, were available for analysis. Mean creatinine was 0.87 mg/dl, mean GFR was 87 ml/min per 1.73 m 2 , and mean creatinine clearance was 100 ml/min. The multivariable-adjusted heritability estimates for creatinine, GFR, and CRCL were 0.29, 0.33, and 0.46, respectively. The peak log of the odds ratio (LOD) scores for serum creatinine, GFR, and CRCL were 2.28 at 176 cM on chromosome 4, 2.19 at 78 cM on chromosome 4, and 1.91 at 103 cM on chromosome 3, respectively. In a community-based sample, measures of serum creatinine, GFR, and CRCL are heritable, suggesting an underlying genetic component. These results also provide suggestive evidence for linkage to measures of kidney function. Further research is necessary to identify the genes involved in the development of kidney disease and to understand their roles in this complex process.

show abstract

Improving the Power of Sib Pair Quantitative Trait Loci Detection by Phenotype Winsorization

Cited by 33 publications

References 16 publications

Effect of Box-Cox Transformation on Power of Haseman-Elston and Maximum-Likelihood Variance Components Tests to Detect Quantitative Trait Loci

Effect of Box-Cox Transformation on Power of Haseman-Elston and Maximum-Likelihood Variance Components Tests to Detect Quantitative Trait Loci

A genome-wide linkage scan identifies multiple chromosomal regions influencing serum lipid levels in the population on the Samoan islands

Genomewide Linkage Analysis to Serum Creatinine, GFR, and Creatinine Clearance in a Community-Based Population

Contact Info

Product

Resources

About