by Sean Downing, Ph.D., MBA, Chief Scientific Officer, Vector Laboratories
Life science researchers have access to more sophisticated tools than ever before. Advances in molecular biology, imaging, analytical technologies, and bioinformatics have dramatically expanded what researchers can measure and characterize. Yet greater technological capability does not necessarily translate into better scientific understanding. In some cases, the abundance of available tools can encourage researchers to focus on what a technology can measure rather than what they fundamentally need to understand about the biology. Even a technically sophisticated experiment can generate inconclusive or misleading results if the underlying biological question is poorly defined.
For researchers, the challenge is therefore not simply identifying the most advanced technologies or highest-performing reagents. It is understanding how underlying biology, experimental approaches, and enabling technologies fit together to produce meaningful evidence. That requires starting with the scientific question and working backward to determine which tools are best suited to answering it.
Defining the question before selecting the tools
One of the most consequential decisions in experimental design happens before samples are collected or reagents are ordered: determining exactly what the experiment needs to establish. Consider a researcher investigating the role of a particular protein in disease progression. The initial instinct might be to identify an antibody against that protein and develop an assay to measure its abundance. But expression levels alone may not provide the information needed to understand its biological function; instead, the most relevant questions could concern where the protein is localized, whether it undergoes a particular post-translational modification, how it interacts with other molecules, or whether its activity changes under specific physiological conditions. Each question may require a different experimental approach.
The same principle applies to biomarkers. Identifying a molecule associated with a disease is not the same as identifying a biomarker that can distinguish between biologically or clinically meaningful states directly related to pathology. Researchers first need to establish what they are trying to differentiate and what evidence would allow them to do so.
Glycobiology provides a useful example. Glycosylation can influence protein folding, stability, molecular interactions, and antibody recognition, yet these modifications may be overlooked in conventional protein detection workflows. PD-L1, for example, is heavily glycosylated in its extracellular domains, including regions recognized by some companion diagnostic antibodies. Research has shown that these glycans can interfere with antibody recognition and that enzymatic deglycosylation can improve detection with certain antibody clones, although the effect is assay dependent.1 Similar glycan-dependent effects on antibody recognition have been described for HER2 and MUC1.2,3
The broader point is that a measurement can be technically accurate while still providing an incomplete picture of the biology. A negative signal does not necessarily mean a target is absent, just as detecting a protein does not by itself tell us whether it is functionally active. Rather than automatically reaching for a more sensitive or sophisticated technology, researchers should first ask what biological state they need to characterize and which molecular features are truly relevant to it. In recent work presented at AACR 2026, our team paired an anti-GFAP antibody with lectins in an in situ proximity ligation assay on human brain tissue. Conventional immunohistochemistry detected GFAP in both normal brain and anaplastic astrocytoma, but the signal for sialylated glycans in proximity to GFAP was largely absent in normal tissue and pronounced in the tumor. The antibody reported the same protein in both samples; the glycan context surrounding it was different.4 Had our team solely focused on the presence vs. absence of GFAP, we would have missed a relevant dimension of the question at hand.
Designing for reproducibility and biological relevance
This alignment between question and method is also fundamental to reproducibility. Antibodies illustrate why. Selecting an antibody based primarily on stated target specificity or reported performance can introduce uncertainty because antibody behavior is highly context dependent. Performance in one application is not a reliable predictor of performance in another, and factors such as sample preparation, antigen accessibility, epitope modification, and detection conditions can all influence results. Reagent selection and validation should therefore be considered part of experimental design rather than procedural steps that occur after the approach has already been established.
The same principle applies when selecting analytical technologies. Different platforms offer different advantages in sensitivity, specificity, spatial information, and quantitative performance, but no technology is inherently superior for every biological question. As experiments become increasingly complex and multiplexed, researchers have more opportunities to generate rich datasets, but also more opportunities for signal interference, analytical variability, or measurements whose biological relevance is uncertain. More data do not guarantee more insight.
A science-first approach establishes what constitutes a meaningful result before the platform is selected, including the controls needed to distinguish genuine biological observations from technical artifacts. Unexpected findings will always be part of research, but thoughtful experimental design makes it easier to determine which unexpected results warrant further investigation.
From reagent suppliers to scientific collaborators
Achieving this alignment between scientific question and experimental approach increasingly requires expertise that extends beyond an individual research team. Modern research brings together biology, chemistry, analytical science, and other disciplines, and important experimental decisions often sit at the intersection of those areas. That creates an opportunity to rethink the relationship between researchers and the companies that develop their tools. Historically, life science suppliers have often operated through a relatively transactional model: a researcher identifies a target, searches a catalog, selects a reagent, and contacts technical support if something goes wrong. While that model provides access to essential research materials, it does little to help determine whether those materials are appropriate for the scientific question. A science-first model instead positions life science companies as collaborators capable of helping researchers solve biological problems rather than simply supplying reagents.
Instead of beginning with which antibody a researcher wants to purchase, a scientific partner can ask what they are trying to understand about the target. Are they investigating expression, localization, molecular interactions, or functional activity? What biological samples are involved? Could sample preparation or post-translational modifications affect detection? What evidence is necessary to support the intended conclusion? The answers may point toward a conventional reagent, a customized assay, or a different experimental strategy altogether. In some cases, the most valuable contribution a scientific partner can make is identifying why the originally proposed approach is unlikely to answer the researcher’s question.
Collaboration should also extend beyond initial assay design. Unexpected results or discrepancies between methods may reflect technical limitations, but they may also reveal biology that was not originally considered. Bringing researchers and scientific experts together to interpret those findings can help determine which is which and guide the next experiment. Getting these decisions right early also accelerates discovery: fewer experiments need to be repeated, unsuitable approaches are identified sooner, and go/no-go decisions rest on evidence the team can trust.
Advancing with a science-first perspective
Ultimately, researchers can apply this same science-first philosophy before beginning any study. Does the proposed measurement directly address the biological hypothesis? Are the reagents appropriate for the relevant biological context? Will the controls distinguish genuine biological effects from technical variability? And, perhaps most importantly, what will the results actually allow us to conclude? Addressing those questions early may require more thought before an experiment begins, but it can prevent considerably more time from being spent troubleshooting an approach that was never well suited to the scientific objective.
As research technologies continue to evolve, it will become increasingly easy to generate more measurements with greater analytical sophistication. But the value of an experiment is not determined by the amount of data it produces or the novelty of the platform used to generate them. It is determined by whether those data advance our understanding of the biological system being studied. Better reagents and more powerful technologies will continue to expand what researchers can investigate, but research progress still depends on starting with the biology, defining the question clearly, and then choosing the tools and expertise needed to answer it.
About the author
Sean Downing, PhD, MBA, is Chief Scientific Officer at Vector Laboratories, where he leads scientific strategy, R&D, and innovation across the company’s US and UK operations as a member of the executive leadership team. He brings more than 25 years of experience across biotechnology, diagnostics, contract research organizations, and biopharmaceutical R&D, with deep expertise in biomarker strategy, precision medicine, and companion diagnostics. Downing trained at Dana-Farber Cancer Institute and Harvard Medical School, is an inventor on multiple US patents, and has published in journals including Nature Medicine and Nature Biotechnology.