Taylor Maxwell

Taylor Maxwell

Associate Research Professor
Faculty: Full-Time
Address: Innovation Hall

Areas of Expertise

Human Disease Genetics, Quantitative Genetics, Statistical Genetics, Population Genetics

• 2014-Present: Associate Research Professor, Computational Biology Institute, George Washington University
• 2007-2013: Assistant Professor, Human Genetics Center, University of Texas School of Public Health

Current Research


The goal of my current research is to dissect the complex genetic architecture of various common chronic diseases such as coronary heart disease (CHD and Alzheimer’s disease and related biological risk factors such as lipids, glucose, and beta amyloid.  In the past I focused on how evolutionary processes and history can be used to direct and inform genotype-phenotype association studies at the local level.  This history at the local level effects all association studies regardless of whether the study is focusing on candidate genes, genome-wide snp arrays (GWAS), or whole genome sequencing.  My current focus has been on context dependent effects such as gene-by-gene (GxG) interactions as well as pleiotropy (where a locus affects multiple traits).  There are many computational, statistical, and interpretation issues with GxG.  Without a priori hypotheses, the number of possible tests creates computational issues and terrible multiple testing issue.  One approach I have focused on is to identify single locus patterns that are likely the result of an underlying GxG or gene-by-environment (GxG) interaction.  I have devoted my time to the methodological developments and empirical application of methods to detect two related single locus patterns, relationship loci (rQTL) and variance heterogeneity loci (vQTL).

Relationship Loci (rQTL)

One problem in evolutionary biology is how a complex multi-trait system with pleiotropy can evolve when a beneficial mutation for one trait may have detrimental consequences for another.  Work by James Cheverud and others have shown that relationship loci (rQTL) create variation in pleiotropy that can be selected upon to further couple or uncouple trait variation and allow joint or separate evolution to occur in complex systems.  Relationship loci (rQTL) occur when the correlation between multiple traits varies by genotype.  They are the product of differential GxG or gene-by-environment interactions.  Differential epistasis occurs when the pattern of interaction between two loci is different for multiple traits.  This creates variation in the pleiotropic effects of a single locus. It also creates a pattern where the correlation between two traits varies by genotype at the single locus level (rQTL).  Beyond the many interesting evolutionary questions that can be addressed with rQTL, they are practical way to identify loci likely to be involved in GxG.  Instead all pair-wise tests, a genome-wide screen would only require the number of markers sampled (i.e. the same as a typical GWAS).  Any significant rQTL presents a prior hypothesis for GxG with one of both of the traits.  The other interacting partner(s) can be identified with a GxG screen where the rQTL is fixed as one of the two loci.  Through rQTL analyses we can identify loci that conditionally affect relationships between known risk factors and CHD, lead to interacting loci, and jointly uncover the relationships between multiple genes and pathways that connect the traits to disease.  We have and are currently empirically identifying rQTL for pairs of lipid traits and subsequently the effect on CHD and upon biomarkers for Alzheimer’s Disease.

Variance Heterogeneity Loci (vQTL)

A variance heterogeneity locus (vQTL) exists when the variance of a quantitative trait varies across the genotypes of a locus.  This pattern can exist for a few different reasons such as particular linkage disequilibrium (LD) and mean effect combinations, biological disruption and context dependencies such as GxG or GxE.  We have done theoretical work to understand the nature of these different factors and have developed and published statistical methodologies to leverage both mean and variance heterogeneity effects to find vQTL.  We are continuing work on other methodological developments and applying our approaches to empirical data related to lipids and CHD and intermediate biomarkers for Azlheimer’s Disease.  vQTL and rQTL are related to each other in interesting ways.  Sometimes a vQTL for a composite trait is a rQTL for the underlying subtraits.  Also, sometimes an rQTL for two traits occurs because it is a vQTL for one of the two traits.  This leads to many a priori hypotheses beyond the original genome-wide screens.


B.S. Brigham Young University, 2000 (Zoology)
Ph.D. Washington University in St. Louis, 2006 (Biology & Biomedical Sciences)


Cao, Y., Maxwell, T.J. and Wei, P., 2015. A Family‐Based Joint Test for Mean and Variance Heterogeneity for Quantitative Traits. Annals of human genetics, 79(1), pp.46-56.

Majithia, A.R., Flannick, J., Shahinian, P., Guo, M., Bray, M.A., Fontanillas, P., Gabriel, S.B., JHS, N., Rosen, E.D., Altshuler, D. and Manning, A.K., 2014. Rare variants in PPARG with decreased activity in adipocyte differentiation are associated with increased risk of type 2 diabetes. Proceedings of the National Academy of Sciences, 111(36), pp.13127-13132.

Ridge, P.G., Maxwell, T.J., Foutz, S.J., Bailey, M.H., Corcoran, C.D., Tschanz, J.T., Norton, M.C., Munger, R.G., O’Brien, E., Kerber, R.A. and Cawthon, R.M., 2014. Mitochondrial genomic variation associated with higher mitochondrial copy number: the Cache County Study on Memory Health and Aging. BMC bioinformatics, 15(7), p.1.

Scientific Journals

Cao Y, Wei P, Bailey M, Kuawe JSK, Maxwell TJ. 2014. A versatile omnibus test for detecting mean and variance heterogeneity. Genetic Epidemiology 38: 51-59.

Maxwell TJ, Ballantyne CM, Cheverud JM, Guild CS, Ndumele CE, Boerwinkle E. 2013. APOE modulates the correlation between Triglycerides, Cholesterol, and CHD through pleiotropy and gene-by-gene interactions.  Genetics 195: 1397-1405.