Multivariate methods and software for association mapping in dose‐response genome‐wide association studies

BioData Mining

Chad C Brown³  Tammy M Havener²  Marisa Wong Medina¹  Ronald M Krauss¹  Howard L McLeod²  Alison A Motsinger‐Reif³ 

[1] , Children’s Hospital Oakland Research Institute, Oakland, CA 94609, USA

[2] Institute for Pharmacogenomics and Individualized Therapy, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA

[3] Bioinformatics Research Center, North Carolina State University, Raleigh, NC, USA

关键词: Simulation study;    GWAS;    MANOVA;    Idarubicin;    Temozolomide;    Chemotherapy;    Chemosensitivity;    Lymphoblastoid cell lines;    Pharmacogenomics;   

Others  :  797224
DOI  :  10.1186/1756-0381-5-21

 received in 2012-04-03, accepted in 2012-12-04,  发布年份 2012

【 摘 要 】

Background

The large sample sizes, freedom of ethical restrictions and ease of repeated measurements make cytotoxicity assays of immortalized lymphoblastoid cell lines a powerful new in vitro method in pharmacogenomics research. However, previous studies may have over‐simplified the complex differences in dose‐response profiles between genotypes, resulting in a loss of power.

Methods

The current study investigates four previously studied methods, plus one new method based on a multivariate analysis of variance (MANOVA) design. A simulation study was performed using differences in cancer drug response between genotypes for biologically meaningful loci. These loci also showed significance in separate genome‐wide association studies. This manuscript builds upon a previous study, where differences in dose‐response curves between genotypes were constructed using the hill slope equation.

Conclusion

Overall, MANOVA was found to be the most powerful method for detecting real signals, and was also the most robust method for detection using alternatives generated with the previous simulation study. This method is also attractive because test statistics follow their expected distributions under the null hypothesis for both simulated and real data. The success of this method inspired the creation of the software program MAGWAS. MAGWAS is a computationally efficient, user‐friendly, open source software tool that works on most platforms and performs GWASs for individuals having multivariate responses using standard file formats.

【 授权许可】

   
2012 Brown et al.; licensee BioMed Central Ltd.

【 图 表 】

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 8.

Figure 10.

【 参考文献 】

• [1]Wheeler H, Dolan M: Lymphoblastoid cell lines in pharmacogenomic discovery and clinical translation. Pharmacogenomics 2012, 13:55-70.
• [2]Welsh M, Mangravite L, Medina M, Tantisira K, Zhang W, Huang R, McLeod H, Dolan M: Pharmacogenomic discovery using cell‐based models. Pharmacological rev 2009, 61(4):413-429.
• [3]Watters JW, Kraja A, Meucci MA, Province MA, McLeod HL: Genome‐wide discovery of loci influencing chemotherapy cytotoxicity. Proc Natl Acad Sci USA 2004, 101:11809-11814.
• [4]Huang R, Duan S, Bleibel W, Kistner E, Zhang W, Clark T, Chen T, Schweitzer A, Blume J, Cox N: A genome‐wide approach to identify genetic variants that contribute to etoposide‐induced cytotoxicity. Proc National Acad Sci 2007, 104(23):9758.
• [5]Dolan M, Newbold K, Nagasubramanian R, Wu X, Ratain M, Cook E, Badner J: Heritability and linkage analysis of sensitivity to cisplatin‐induced cytotoxicity. Cancer res 2004, 64(12):4353.
• [6]Bleibel W, Duan S, Huang R, Kistner E, Shukla S, Wu X, Badner J, Dolan M: Identification of genomic regions contributing to etoposide‐induced cytotoxicity. Human genet 2009, 125(2):173-180.
• [7]Watson VG, Motsinger‐Reif A, Hardison NE, Peters EJ, Havener TM, Everitt L, Auman JT, Comins DL, McLeod HL: Identification and Replication of Loci Involved in Camptothecin‐Induced Cytotoxicity Using CEPH Pedigrees. PLoS ONE 2011, 6:e17561.
• [8]Peters EJ, Kraja AT, Lin SJ, Yen‐Revollo JL, Marsh S, Province MA, McLeod HL: Association of thymidylate synthase variants with 5‐fluorouracil cytotoxicity. Pharmacogenet Genomics 2009, 19:399-401.
• [9]Peters E, Motsinger‐Reif A, Havener T, Everitt L, Hardison N, Watson V, Wagner M, Richards K, Province M, McLeod H: Pharmacogenomic characterization of US FDA‐approved cytotoxic drugs. Pharmacogenomics 2011, 12(10):1407-1415.
• [10]Watson V, Hardison N, Harris T, Motsinger‐Reif A, McLeod H: Genomic Profiling in CEPH cell lines distinguishes between the Camptothecins and Indenoisoquinolines. Mol Cancer Ther 2011, 10(10):1839-1845.
• [11]Duan S, Huang R, Zhang W, Mi S, Bleibel W, Kistner E, Cox N, Dolan M: Expression and alternative splicing of folate pathway genes in HapMap lymphoblastoid cell lines. Pharmacogenomics 2009, 10(4):549-563.
• [12]Gamazon E, Duan S, Zhang W, Huang R, Kistner E, Dolan M, Cox N: PACdb: a database for cell‐based pharmacogenomics. Pharmacogenetics and genomics 2010, 20(4):269.
• [13]Stark AL, Zhang W, Mi S, Duan S, O’Donnell PH, Huang RS, Dolan ME: Heritable and non‐genetic factors as variables of pharmacologic phenotypes in lymphoblastoid cell lines. Pharmacogenomics J 2010, 10:505-512.
• [14]Brown C, Havener T, Everitt L, McLeod H, Motsinger‐Reif A: A comparison of association methods for cytotoxicity mapping in pharmacogenomics. Frontiers in Stat Genet And Methodology 2012, 2:1-14.
• [15]Brown CC, Havener TM, Medina MW, Mangravite L, Kraus RM, McLeod H, Motsinger‐Reif A: A genome‐wide association analysis of temozolomide response using lymphoblastoid cell lines reveals a clinically relevant association with MGMT. Pharmacogenetics and Genomics 2012, 22(11):796-802.
• [16]Hegi M, Liu L, Herman J, Stupp R, Wick W, Weller M, Mehta M, Gilbert M: Correlation of O6‐methylguanine methyltransferase (MGMT) promoter methylation with clinical outcomes in glioblastoma and clinical strategies to modulate MGMT activity. J Clinical Oncol 2008, 26(25):4189-4199.
• [17]Schmidt W, Kalipciyan M, Dornstauder E, Rizovski B, Steger G, Sedivy R, Mueller M, Mader R: Dissecting progressive stages of 5‐fluorouracil resistance in vitro using RNA expression profiling. Int j cancer 2004, 112(2):200-212.
• [18]Jarek S: mvnormtest: Normality test for multivariate variables. 2009. [R package version 0.1‐7]
• [19]Doornik J, Hansen H: An Omnibus Test for Univariate and Multivariate Normality*. Oxford Bulletin of Economics and Stat 2008, 70:927-939.
• [20]Timm N: Applied multivariate analysis. 175 Fifth Avenue, New York, NY 10010, USA: Springer Verlag New York, Inc.; 2002.
• [21]Brown C, Motsinger‐Reif A: Software for genome‐wide association studies having multivariate responses: Introducing MAGWAS. NC State Department Stat Tech R 2012, 1-9.
• [22]Purcell S, Neale B, Todd‐Brown K, Thomas L, Ferreira M, Bender D, Maller J, Sklar P, De Bakker P, Daly M: PLINK: a tool set for whole‐genome association and population‐based linkage analyses. The Am J Human Genet 2007, 81(3):559-575.
• [23]Purcell S: PLINK v1.07. 2009. [Http://pngu.mgh.harvard.edu/purcell/plink/ webcite]
• [24]Hicklin J, Moler C, Webb P, Boisvert RF, Miller B, Pozo R, Remington K: JAMA v1.0.2. 2011. [Http://math.nist.gov/javanumerics/jama/ webcite]
• [25]Steinmetz PN, Warnes G, Warnes J: DistLib v0.9.1. 2007. [Http://statdistlib.sourceforge.net/ webcite]
• [26]Molenberghs G, Verbeke G: Models for discrete longitudinal data. 233 Spring Street, New York, NY 10013, USA: Springer Science+Business Media, Inc.; 2006.
• [27]Wu S, Fu G, Chen Y, Wang Z, Wu R: Genetic mapping of complex traits by minimizing integrated square errors. BMC genet 2012, 13:20.
• [28]Wu S, Yang J, Wu R: Semiparametric functional mapping of quantitative trait loci governing long‐term HIV dynamics. Bioinformatics 2007, 23(13):i569—i576.
• [29]Yang J, Wu R, Casella G: Nonparametric functional mapping of quantitative trait loci. Biometrics 2008, 65:30-39.