Higher Resolution of Exposome-by-Genome Interactions: unifying genomic and exposomic science in search of relevant phenotypic variation

Last month, researchers from genomics, epidemiology, environmental health, and computational biology met at the Mendel Museum in Brno—quite literally in the garden where Gregor Mendel first uncovered the rules of inheritance.

Gathering in this symbolic space, we explored the future of exposome-by-genome science and asked how our community can build next-generation tools for understanding human variation.

What emerged was a unified message: the time has come to bring the rigor, scale, and structure of genomics to the exposome and execute on ExWASs: conducting exposure-wide analyses in the same spirit that re-focused the genetics community¹.

Mendel: A reminder that inheritance has two halves

Genomics has transformed biomedical science because it developed standards: reference genomes, coordinated pipelines, shared cohorts, transparent methods. Exposomics—the study of environmental exposures and their biological impact—has lacked comparable structure.

At the meeting, new scientific findings highlighted why we need that structure. For example, work discussed by Kári Stefánsson showed that even canonical genetic loci such as FADS1 behave very differently depending on ancestry, geography, and diet. Longitudinal proteomic analyses, such as those applied to the FOURIER trial, also demonstrated that protein signatures change with therapy and environmental influence.

This is the exposome in action: dynamic, context-dependent, and biologically meaningful.

Moving beyond the “one exposure at a time” paradigm

The study of genetic variation is a study of evolution and how historical exposures and events impose pressure on the genome to change through generations. Many foundational quantitative-genetics models treat exposures as isolated and independent factors, studied rigorously through experimentation. But today’s exposome is:

• measured observationally
• multi-dimensional
• temporally dynamic
• correlated across behaviors, geography, and policy
• and sometimes reversible.

This complexity demands a new generation of analytic frameworks—exposomic reference matrices, longitudinal designs, unified metadata standards, and causal models that consider interactions rather than isolated effects to tease apart disease etiology. As it should be, the phenotype is front and center.

Where is the missing variation?

Multiple modern studies point to substantial unexplained phenotypic variation after accounting for both the genome and the limited, a priori–selected set of exposures traditionally measured (^{2 3}). This raises a central question for disease researchers (be it exposomics, genomics, or disease researchers): where is the rest of the phenomic and disease variation hiding?

A major theme emerging from the workshop was the need to expand beyond candidate exposures and to invest in data-driven discovery frameworks including exposure-wide association studies that systematically search for environmental and behavioral contributors to disease outside what is explained by “heritability” alone. In other words, if genetics explains only part of disease risk, we must build technologies, assays, and analytic pipelines capable of uncovering the environmental architecture of the remainder.

Participants also noted that the strict definitions of G×E used in statistical genetics (e.g., as discussed by ⁴) are important, but can be limiting. While multiplicative interaction terms are important, they represent only one corner of the broader landscape of genome–environment interplay. Many meaningful interactions may instead manifest as context-dependent main effects, threshold shifts, buffering or decanalization phenomena, or additive modifications that alter risk without classic statistical “interaction”.

Expanding our conceptual framework will be critical to capturing the full spectrum of variation that neither genetics nor narrowly defined exposures can currently explain.

The role of ecosystems, time, and intergenerational processes

Discussions led by genomics pioneers emphasized that exposures must be understood not only at the individual level but also at ecosystem and reproductive scales. Environmental influences act across tissues and, in some cases, across generations. Increasingly, EMBL-EBI resources—such as the GWAS Catalog, metabolomics repositories, and exposome ontologies—are emerging as crucial infrastructure for this work.

A key takeaway: time is a missing variable in most exposure studies, and must be incorporated into next-generation exposomics.

From cohorts to a “reference exposome”

Speakers from ALSPAC, RECETOX, and other large cohorts—including Nick Timpson, Jana Klanová, and Adam Ledanowski—pointed toward a practical strategy:

…extend existing genetic cohorts with high-quality exposomic data rather than building everything from scratch.

Discussions focused on:

• the potential for population-scale proteomics and MRI phenotyping,
• efficient sampling strategies (as highlighted by Nilanjan Chatterjee),
• and stitching fragmented datasets into unified, interoperable resources.

These efforts point toward the long-term goal of a reference exposome: a structured, population-level baseline akin to the reference genome.

Gene–environment interactions: what impactful use cases teach us

Chatterjee’s work on tobacco-related cancer provided an important real-world example of gene–environment interaction with population-wide impact in an highly exposed population in India. Tobacco exposure has a dominant main effect, but genetic variation modulates susceptibility in ways that influence screening and prevention. This is G×E at its most useful: a strong environmental factor whose effects are sharpened by understanding genetic background. Perhaps there are lessons here in dissecting diabetes etiology around the world, where obesity is a well-known causal factor whose “penetrance” is variable around the world – in individuals with different genetic ancestry.

On the other end of the spectrum, Shamil Sunyaev discussed incomplete penetrance as a form of G×E. Even variants long thought to carry high risk show substantial variation in expression depending on environmental, epigenetic, or developmental context. What looks like “noise” in penetrance is often an environmental signal. Furthermore, families and disease inheritance – another data substrate for genomics investigation – can tell us about GxE.

Together, these perspectives show that G×E spans a continuum, from exposure-driven outcomes modulated by genotype to genotype-driven outcomes modulated by environment.

The architecture of interaction: indirect pathways, buffering, and breakdown

Greg Gibson expanded the discussion into biological networks, emphasizing indirect pathways such as neurohormonal signaling and glucocorticoid response. His work on decanalization—the loss of phenotypic stability under environmental stress—offers a mechanistic explanation for how exposures can unmask latent genetic variation.

These ideas reinforce a central theme: to explain human variation, we must measure environmental structure with the same depth and rigor applied to the genome. Please read his blog post and discussion with the community here: https://genomestake.substack.com/p/exposing-and-finding-the-heritability

What the community needs: standards, coordination, and shared infrastructure

Across breakout sessions, attendees converged on a roadmap familiar to anyone who has lived through the rise of genomics:

1. Metadata and measurement standards across mass spectrometry, sensors, geospatial data, and cohort studies.
2. Cross-scale environmental repositories, including human, model organism, and ecosystem samples.
3. Ethical and governance frameworks for sensitive environmental and health data.
4. Longitudinal exposure measurement, capturing timing, reversibility, and mediation.
5. Interoperable tools and ontologies so that exposomic data can integrate seamlessly with genomic and clinical data.

These themes echo our reflections from the 2025 Gordon Research Conference: exposomics will require bringing communities together—environmental scientists, geneticists, epidemiologists, methodologists, engineers, and clinicians.

Why now? The case for a unified science of variation

The exposome is the modifiable half of human biology. It is where prevention, policy, and molecular science intersect. But without structured frameworks, its potential remains fragmented.

The Mendel Museum reminded us that variation arises from rules—but rules shaped by both genome and environment. To understand disease, resilience, and aging at scale, we must treat these two domains as inseparable parts of a single system.

Standing in Mendel’s garden, the path forward was clear: The next era of biomedical science will be exposome-by-genome science.

We now have the tools, cohorts, and computational frameworks to build it—and the community momentum to make it real.

lease check out an interview (Gary Miller and Chirag Patel) on Mendelspod with the fantastic Theral Timpson here: https://www.mendelspod.com/p/from-gwas-to-ewas-chirag-patel-and