The molecule that carries the message
Science has a translation problem. Not in the linguistic sense, but in the deeper sense of passage: the long, uncertain journey a discovery must make from a well-characterized biological signal to a clinical decision, a therapeutic strategy, or a public health intervention. Most molecules that enter that journey do not complete it. The distance between what we can measure and what we can act on remains one of the most consequential gaps in modern biomedicine.
I have spent over a decade working at the intersection of metabolomics and multiomics, and with the scientific communities that connect them. The question that drives me, more than any other, is this: what would it take to close that gap? Not in a single study, but structurally, at the level of how fields organize themselves and how knowledge actually moves from discovery to application.
The answer, I have come to believe, involves metabolomics at its center. Not simply as a technology, but as a biological and strategic bridge.
How multiomics integration has changed the way we explore biology
For much of modern biological research, the dominant model was reductionist by necessity. You chose a system, a molecule class, a pathway, and you went deep. That depth produced extraordinary science. It also produced silos: datasets that could not speak to each other, and biological conclusions that turned out to be context-dependent in ways investigators had not anticipated. For example, patients with similar genomic profiles may still exhibit dramatically different responses to therapy or disease progression.
Multiomics integration has disrupted that model, not by replacing depth, but by adding a dimension that reductionism could not provide. When genomic, transcriptomic, proteomic, and metabolomic data are aligned around a shared biological question, something emerges. Biological mechanisms that once appeared linear reveal layers of feedback, compensation, and cross-system interaction. Phenotypes that seemed straightforward turn out to be subtypes with distinct biological drivers.
What has fundamentally changed is the way we frame biological questions. We are no longer asking what a gene does or what a protein signals in isolation. We are asking how a biological system behaves under disease, under treatment, across populations. That is a fundamentally different question, and it demands fundamentally different infrastructure: analytical, computational, and human.
Metabolomics as the interpretive bridge
Within integrated multiomics frameworks, metabolomics captures functional output – the net result of all upstream influences expressed as a biochemical phenotype. It is the layer closest to clinical observables: symptoms, treatment response, disease progression, and physiological adaptation. Metabolites reflect the cumulative influence of genetics, environment, microbiome activity, lifestyle, and disease state in real time.
This proximity to clinical reality is what makes metabolomics so powerful as a translational tool and so demanding to interpret well. When metabolism changes, biology is responding to something real and functionally important. But importance is not the same as actionability. Moving from a significant metabolic finding to a validated biomarker or clinical decision support tool requires not just better data, but better frameworks for interpretation. Those that integrate biological knowledge, clinical expertise, regulatory understanding, and computational sophistication simultaneously. That convergence does not happen automatically. It has to be deliberately built by communities structured to enable it.
The translation gap and what it will take to close it
The gap between discovery and application in metabolomics is not primarily a data problem. We are generating data at scale. The bottleneck lies downstream: in our collective capacity to interpret findings in biologically rigorous and clinically relevant ways, and to communicate across the disciplinary boundaries that separate the people who generate evidence from those who apply it.
Two things are needed, and they are deeply connected. The first is education: not narrow technical training, but genuine scientific fluency across the multiomics ecosystem. Researchers who understand both the power and limitations of each data layer. Clinicians who can engage meaningfully with metabolomics outputs. Computational scientists who understand the biological context well enough to ask meaningful questions. These are not unreasonable expectations, but they require sustained, intentional investment in how we train people and how we design the environments they work in.
The second is structural: better interfaces between the sectors that metabolomics must traverse to deliver clinical impact. Academic discovery, industry application, clinical validation, and regulatory translation currently operate with insufficient shared language, insufficient data standards, and insufficient mechanisms for the cross-sector dialogue that translational science actually requires. Building those interfaces is as much a scientific priority as building better instruments. Some of the most productive scientific conversations I’ve seen happen when analytical scientists, clinicians, computational biologists, and industry partners are able to challenge each other’s assumptions productively.
Why diverse perspectives are a scientific imperative
The complexity of multiomics interpretation means that the composition of the teams doing that work is in itself a scientific variable. Every analytical decision, such as which features to prioritize, which biological assumptions to apply, and which clinical comparators to use, shapes interpretation. Teams drawing from narrower scientific perspectives are more likely to overlook biologically meaningful signals.
Cross-sector diversity matters here, too. The scientist who has spent a decade in a clinical setting reads a metabolomics dataset differently than one who has spent it in an analytical laboratory. Both readings are necessary. The translation gap closes fastest when those perspectives are in genuine dialogue, structured not around deference to a dominant methodology, but around a shared commitment to biological truth.
Where the greatest opportunity lies
The next five years in multiomics will be defined less by what we can measure and more by what we can understand and act upon. The platforms are advancing. AI-assisted integration is beginning to help researchers navigate the scale and complexity of multiomic datasets in ways that were previously difficult to operationalize.
But the transformative opportunity lies in building the translational infrastructure that allows those advances to reach patients and populations: federated data ecosystems in which metabolomics signatures can be validated against clinical outcomes at scale; standardized reporting frameworks that make findings reproducible across institutions; and cross-sector partnerships structured around the translation problem, not just the discovery problem. Metabolomics is the layer of multiomics closest to biological and clinical reality. The question is whether the communities around it are organized to take full advantage of that proximity.
What does “From Molecules to Medicine” mean to you?
It means the distance between those two words is the whole of the problem and the whole of the opportunity—molecules we can measure with extraordinary precision. Medicine requires something more: consensus, validation, communication, and the kind of cross-disciplinary trust that only communities can build over time. That journey is what motivates me. Not the data alone, but what it might become when the right people are in genuine conversation around it.
What would you tell women pursuing interdisciplinary science?
Develop fluency across boundaries – not just technical ones, but cultural ones. Learn how clinicians think. Learn how regulators evaluate evidence. Understand what your computational colleagues are worried about when they look at a complex dataset. The scientists who will have the greatest impact in the next decade of multiomics are not necessarily the deepest specialists but those who can move across disciplines without losing rigor and who understand that translation is itself a scientific skill worth cultivating deliberately.
