At the 2021 Advances in Genome Biology and Technology meeting, John Greally had some unusual advice with respect to confounders in epigenomic studies.

Epigenomics, he told the audience in his talk on "Thinking beyond the creode: epigenomics and human disease," has real promise for understanding genomic mechanisms of disease. "But possibly not in the way we think."

The expression "creode" was the first word used to define the developmental path taken by different cells, and explain how radically different cells could come from the same source during embryonic development. Biologist C.H. Waddington, who first coined the term, was trying to explain the influence of genetic variation on cell state.

Today, a major explanation for variations in cell state is that epigenetic marks affect how well the transcription machinery can get to a given bit of DNA. Those accessibility differences affect which genes are expressed in a cell, and consequently, which cell type it is. Experience can change the epigenetic makeup of a cell, and the epigenetic marks can be inherited, which would account for multigenerational effects of environmental exposures.

Methylation, though, which is a frequently investigated epigenetic marker, may be far more dynamic than this view allows for. Greally said that some epigenetic markers such as polycomb and heterochromatin changes appear to be much more stable than methylation, and could be the specific epigenetic inheritance mechanisms.

And epigenetic changes do not necessarily mean that any individual cell has changed its state -- that cells were "sitting in Waddington's creode, and now epigenetics has caused them to sit in another creode," he said.

Greally talked about changes in cell subtype proportion, transcription factor binding, and DNA polymorphisms as possible confounders that can cause, rather than be caused by, changes in DNA methylation.

Greally's central argument, though, was that "confounders" are also a source of information, rather than a nuisance.

"We can either try to clean up our data and get rid of all these confounding influences, or, and I would suggest that this is the more productive approach, embrace these confounding influences," said Greally, who is a professor of genetics, medicine and pediatrics at Albert Einstein College of Medicine.

"If [you] find that there are changes in cell subtypes that explain the differences between your cases and controls... you have found something that is relevant to pathogenesis of the condition," he said. The same holds true for changed transcription factor activity, which can give insight into cell signaling pathways, and sequence polymorphisms.

"We should be harvesting these confounders," he said.

He gave examples for each approach -- a study using an epigenomics approach to identify changed immune cell proportions in lupus patients, and other identified transcription factors that could be used to re-create a bone marrow niche for hematopoietic stem cells.

The approach Greally predicted would be "particularly fruitful" is to use epigenomic assays to reveal sequence variation. As an example, he described research published by the Broad Institute's Melina Claussnitzer and her colleagues in the March 2, 2021, print issue of Cell Metabolism.

In their study, the researchers identified a regulatory subunit, rs56371916, that changed the binding affinity of the transcription factor SREBP1. This changed binding affinity specifically impacted the expression of ADCY5, which codes for the metabolic enzyme ADCY5, by "altering the chromatin landscape from poised to repressed." In their work, Claussnitzer and her colleagues showed that one consequence of the variant was that it affected how efficiently progenitor cells differentiated into osteoblasts, specialized cells that break down bone. Those results could explain ADCY5's link to osteoporosis, and serve as an important reminder that a relationship between epigenetics and outcomes does not necessarily mean there is a mediating change in cell fate.

"In the field of epigenetics, we look at things like exposures and behavioral and phenotypic outcomes and molecular outcomes," Greally said. "What we tend to forget to do is to look at the actual tissue."

While rs56371916 is a relatively common variant in East Asian populations, Greally said that where the epigenomic approach may really shine is in understanding the function of ultrarare variants.

The functional effects of rare variants identified in population genomics can be hard to determine, because they are often found in only one individual or one family, and are likely to be heterozygous.

But "multiple rare variants can have the same outcome on DNA methylation or chromatin structure and show allelic effects on one allele versus the other," Greally said, which means that it is possible to use epigenomic assays as a "convergent molecular readout of a lot of different variants... This is potentially the most powerful thing we are going to be able to do with epigenomic assays over the next several years."