Genetic imprinting and maternal genotype effects are two epigenetic phenomenon, which may lead to heritable changes in gene expression or cellular phenotype without altering the underlying DNA sequence. These epigenetic factors, also known as parent-of-origin effects, have been increasingly explored for their roles in complex traits as part of a concerted effort to find the "missing heritability". Research on parent-of-origin effects has taken on a new dimension as the next generation sequencing technology becomes wildly available. However, despite the great biological and technological progress and innovation, statistical methods are lacking behind. Most existing methods for studying imprinting/maternal effects are restricted to data from nuclear families. The applicability of such methods is further hindered by the need to make strong but unrealistic assumptions to reduce the number of parameters to avoid overparametrization. In addition, imprinting and maternal effects are confounded and maternal effect is believed to be heterogeneous; all these further complicate modeling and analysis efforts. This project takes up this challenging problem and aims to develop novel statistical and computational models/methods and software for extended families and nuclear families without making the strong but unrealistic assumptions. For nuclear family data from retrospective studies, in addition to case families, the study design also considers controls, which can be in the form of control nuclear families or internal controls from unaffected siblings of case families. This novel design makes it possible to formulate a partial likelihood approach, wherein the likelihood component of interest is free of the nuisance parameters. This circumvents the problem of overparametrization and unrealistic assumptions that plague existing methods. For extended pedigrees from prospective studies, the focus is on the development of a method for incorporating heterogeneous maternal effects and addressing the issue of missing data, which is common for extended pedigrees. Methods proposed will be implemented in software that will be made publicly available.
The importance of epigenetics in the twenty first century cannot be overstated. To borrow science writer David Shenk's words, epigenetics is ``perhaps the most important discovery in the science of heredity since the gene''. In this regard, genomic imprinting and maternal effects, two aspects of the epigenetic process, holds important roles as they are essential for normal mammalian growth and development but can also cause devastating diseases and birth defects. Genetic imprinting refers to the epigenetic marking of the parental origin of a gene, which leads to the same DNA sequence being expressed differently depending on whether it is inherited from the mother or from the father. It is well known that Prader?Willi syndrome and Angelman syndrome are genetic disorders involving genetic imprinting. Maternal effect refers to the influence of the prenatal environment provided by a mother, which may arise from the expression levels of some genes carried by the child being altered by the additional genetic materials passed from the mother during pregnancy. Biological research increasingly reveals the presence and importance of maternal effect in many diseases such as childhood cancer and birth defects. This projects aims to develop novel statistical methods and computational software that circumvent difficulties in identifying genes that bear imprinting and/or maternal effects using nuclear and extended family data. The tools developed will be made available to the larger scientific community to aid scientific discoveries and finding treatments for genetic diseases. This project will also contribute to the training of the next generation of researchers in a cutting-edge interdisciplinary research area that fuses knowledge in biology, statistics and computer science.
