E. Andrés Houseman
- Published in print:
- 2014
- Published Online:
- December 2014
- ISBN:
- 9780198709022
- eISBN:
- 9780191779619
- Item type:
- chapter
- Publisher:
- Oxford University Press
- DOI:
- 10.1093/acprof:oso/9780198709022.003.0015
- Subject:
- Mathematics, Probability / Statistics, Biostatistics
Deoxyribonucleic acid (DNA) methylation is tightly linked with cellular differentiation. For instance, it has been observed that DNA methylation in tumor cells encodes phenotypic information about ...
More
Deoxyribonucleic acid (DNA) methylation is tightly linked with cellular differentiation. For instance, it has been observed that DNA methylation in tumor cells encodes phenotypic information about the tumor. Thus, understanding of tumor biology is fruitfully enhanced by the study of the multivariate structure of DNA methylation data. To the extent that such data possess discrete latent structure, it can be viewed as encoding different tumor subtypes (in cancer studies) or tissue types (more generally). However, in some cases there may be more evidence of continuous latent structure reflecting a continuous range of variation. This chapter discusses several specific latent variable models that have been used in the last decade to analyze DNA methylation data, including approaches for modeling DNA methylation data in low-dimensional settings such as in candidate gene studies and recursively partitioned mixture model approaches for modeling DNA methylation in high-dimensional settings.Less
Deoxyribonucleic acid (DNA) methylation is tightly linked with cellular differentiation. For instance, it has been observed that DNA methylation in tumor cells encodes phenotypic information about the tumor. Thus, understanding of tumor biology is fruitfully enhanced by the study of the multivariate structure of DNA methylation data. To the extent that such data possess discrete latent structure, it can be viewed as encoding different tumor subtypes (in cancer studies) or tissue types (more generally). However, in some cases there may be more evidence of continuous latent structure reflecting a continuous range of variation. This chapter discusses several specific latent variable models that have been used in the last decade to analyze DNA methylation data, including approaches for modeling DNA methylation data in low-dimensional settings such as in candidate gene studies and recursively partitioned mixture model approaches for modeling DNA methylation in high-dimensional settings.