What Are Biobanks and Why Are They Important for Research?

Biobanks are organized repositories that collect, process, and store human biological samples, such as blood, tissue, saliva, and DNA, along with the health and lifestyle data of the people who donated them. They function as research libraries of human biology, and their importance has grown dramatically over the past two decades as genetics, data science, and personalized medicine have converged. Some of the largest, like the UK Biobank with its half-million participants, have become foundational infrastructure for thousands of studies across virtually every area of medicine.

What Exactly Gets Stored in a Biobank

The word “biobank” covers a range of operations, from a single hospital freezer holding leftover surgical tissue to a national enterprise with millions of samples linked to decades of health records. What they share is a core mission: preserve biological material so researchers can study it later, sometimes years or decades after collection. The samples themselves vary widely. Blood is the most common, often separated into plasma, serum, and white blood cells before storage. Tissue samples come from surgeries or biopsies. Saliva, urine, stool, and even cerebrospinal fluid may also be banked. Increasingly, biobanks store extracted DNA, RNA, and proteins rather than (or in addition to) the raw tissue.

Equally important is the data attached to each sample. A vial of blood sitting alone in a freezer is of limited use. Paired with the donor’s medical history, medication records, imaging results, lifestyle questionnaire answers, and sometimes wearable-device data, that vial becomes a window into how biology and environment interact to produce health or disease. When biobanks are linked to electronic health records, the research possibilities expand further, because participants’ health trajectories update automatically over time without requiring them to return for follow-up visits.¹

How Samples Are Kept Viable for Years

Freezing is the backbone of biobank storage. Most facilities rely on ultra-low-temperature freezers set to around −80°C, which can preserve DNA and protein for years. RNA is more fragile, though, and can degrade at that temperature over about five years unless the sample has been treated with a stabilizing solution before freezing.² Some biobanks go colder, storing samples in the vapor phase of liquid nitrogen at roughly −150°C or below. A comparison of breast-cancer samples stored under both conditions found that RNA quality was significantly better in liquid-nitrogen storage than at −80°C, while DNA quality did not differ meaningfully between the two methods.²

For clinical biobanks that collect tissue during surgery, getting the sample from the operating room to a freezer quickly is critical. Portable dewars filled with liquid nitrogen serve as interim transport containers, and even the choice of cryotube material can affect sample safety during that short transit.³ These details sound mundane, but a sample damaged by a warm spell or a cracked tube can quietly invalidate downstream research.

Not all samples are frozen fresh. Hospitals have archived formalin-fixed, paraffin-embedded (FFPE) tissue blocks for over a century. These waxy blocks preserve tissue structure well but are harsher on nucleic acids. Still, modern sequencing technology has caught up: a study comparing fresh-frozen and FFPE pairs from six human tissue types found that gene-expression profiles correlated strongly between the two preservation methods, even in FFPE samples stored for up to 20 years.⁴ That finding has opened hospital pathology archives to genomic research in a way that was not feasible a decade ago.

The Major Biobanks and How They Differ

Population-based biobanks recruit from the general public and follow participants over time. The UK Biobank enrolled roughly 500,000 people between 2006 and 2010 and has since become one of the most heavily used research datasets in the world. Other large population biobanks include the CONSTANCES project in France, Germany’s National Cohort (NAKO), LifeLines in the Netherlands, FinnGen in Finland, and the All of Us Research Program in the United States.⁵ Each has its own design: some emphasize genetic data, others prioritize environmental exposures, and some focus on specific populations or health systems.

Hospital-based biobanks, by contrast, collect samples from patients during routine clinical care. They tend to have richer clinical detail for specific diseases but less information about healthy volunteers. Both types feed modern research, and the line between them is blurring as hospital biobanks increasingly link to broader health records and population biobanks add imaging and clinical follow-up.

Genomics and the Hunt for Disease-Risk Genes

The single biggest use of biobank data in recent years has been genome-wide association studies, or GWAS. These scan the DNA of hundreds of thousands of people, looking for genetic variants that show up more often in people with a given disease than in those without it. Biobanks make this possible by providing both the genetic material and the health outcomes needed for comparison. A collaborative effort called the Global Biobank Meta-analysis Initiative pools GWAS data across biobanks worldwide to boost the statistical power of these searches, improving the ability to find genetic signals for rarer diseases and to build better risk-prediction tools.⁶

These genetic maps do more than satisfy scientific curiosity. By identifying the genes and biological pathways linked to a disease, researchers can nominate potential drug targets before a single experiment is run in a lab. For complex diseases like type 2 diabetes, coronary artery disease, or depression, where hundreds of small-effect genetic variants contribute, biobank-scale GWAS has been transformative. A recent pan-UK Biobank analysis, for example, used genome-wide data to refine genetic associations across diverse ancestral backgrounds, improving the resolution of findings that had previously been dominated by data from people of European descent.⁷

Predicting How You Will Respond to a Drug

Pharmacogenomics, the study of how your genes affect your response to medications, is one of the most immediately practical applications of biobank research. An analysis of nearly 500,000 UK Biobank participants found that virtually all of them, about 99.5%, carried genetic variants predicted to cause an atypical response to at least one medication. On average, each person had variants linked to an unusual response to about ten different drugs. And nearly a quarter of participants had already been prescribed a drug for which their genetics predicted a non-standard response.⁸

Those numbers come from looking at 14 well-characterized genes known to influence drug metabolism. This is not speculative science: the findings involve drugs people take every day, including blood thinners, statins, antidepressants, and pain relievers. A separate study using Estonia’s biobank, which links genetic data to pharmacy purchase records for over 200,000 people, confirmed strong genetic signals for dose variation in warfarin (a blood thinner), metoprolol (a blood-pressure drug), and commonly prescribed statins. The genetic variants affecting dosing operated independently of disease-risk factors, meaning they represent a separate layer of personalization that current prescribing practices often miss.⁹

Tracking Chronic Disease Over Time

Because biobanks follow participants for years, they are especially valuable for studying chronic diseases, conditions that develop slowly and involve many interacting factors. Cardiovascular disease, diabetes, and cancer all fall into this category, and biobank samples provide molecular-level detail that clinical records alone cannot capture.¹⁰ Researchers can look back at a stored blood sample taken years before a participant developed heart disease and ask what molecular markers were already changing, long before any symptoms appeared.

The UK Biobank’s longitudinal design, with a median follow-up of about six years in early analyses and growing longer every year, has enabled studies connecting everything from grip strength and physical activity to cardiovascular events and mortality in over half a million people.¹¹ Other researchers have used the same dataset to examine how physical activity interacts with multimorbidity, the accumulation of two or more chronic conditions, and its relationship to life expectancy.¹² These studies would be nearly impossible without a biobank’s combination of biological samples, health data, and time.

Pandemic Response and Infectious Disease

COVID-19 demonstrated both the value and the fragility of biobanking infrastructure. When SARS-CoV-2 emerged, researchers urgently needed samples from infected and vaccinated people to develop diagnostic tests, monitor immune responses, and track new variants. Biobanks that already had pre-pandemic samples could compare a person’s biology before and after infection, a study design that cannot be set up retroactively. The speed of access to well-characterized samples proved critical: Canada’s CoVaRR-Net biobank, for instance, provided standardized samples and clinical data that supported research on immune durability and the behavior of new variants of concern.¹³

In Uganda, a biobank that had been built up through earlier epidemic preparedness work pivoted to COVID-19 by providing stored samples for the validation of 17 diagnostic kits. Kits that passed validation were then deployed for mass screening, accelerating the detection and isolation of cases.¹⁴ The lesson was clear: countries that invested in biobanking before a crisis had a measurable head start when one arrived.

Liquid Biopsy and Cancer Monitoring

One of the more exciting developments enabled by biobank research is the liquid biopsy, a blood draw used to detect fragments of tumor DNA circulating in the bloodstream. Biobanks are essential for developing and validating these tests because researchers need large numbers of well-annotated plasma samples paired with confirmed clinical outcomes. In bladder cancer, for instance, researchers used biobanked plasma and urine samples collected over as long as 20 years to develop personalized assays that could detect tumor DNA even in patients with non-invasive disease. Patients whose disease later progressed showed significantly higher levels of circulating tumor DNA in samples taken before that progression was clinically apparent.¹⁵

Getting liquid-biopsy research right depends heavily on how plasma samples are collected and processed. Variations in blood draw technique, the time between collection and centrifugation, and the choice of collection tube can all affect the quality of circulating DNA recovered from a sample.¹⁶ This is where biobank quality-control standards become a direct bottleneck for clinical progress.

Quality Control Is Harder Than It Sounds

Most errors in clinical laboratory work come from pre-analytical steps: everything that happens to a sample before it reaches the testing instrument. The same is true for biobanks. How quickly blood is processed after a draw, whether a tissue sample sat at room temperature before freezing, even the type of collection tube used can alter the molecules researchers later try to measure.¹⁷ If those variables are not controlled or at least recorded, downstream studies can produce misleading results.

Identifying reliable quality-control markers has been a slow process. A review by the International Society for Biological and Environmental Repositories found that only a handful of markers had both a known threshold for detecting pre-analytical damage and a reference range that labs could use. For example, elevated levels of certain growth factors in serum can indicate the sample was exposed to high temperatures or went through freeze-thaw cycles, but the set of well-validated markers remains small.¹⁸ Expanding this toolkit is an active area of work, because as downstream assays become more sensitive, they also become more susceptible to artifacts introduced by sloppy handling.

Machine Learning Meets Biobank Data

The sheer scale of biobank datasets has made them a natural testing ground for machine learning. When you have genetic data, blood biomarkers, imaging, activity-tracker data, and health records for hundreds of thousands of people, traditional statistical methods can struggle to capture the complex interactions between variables. Machine learning models trained on UK Biobank data have shown improvements in disease-risk prediction across multiple conditions. One approach, called survivalFM, improved discrimination (the ability to correctly rank who will get sick) in roughly a third of the disease scenarios tested and improved reclassification of risk, meaning it moved people into more accurate risk categories, in the vast majority of scenarios.¹⁹

Wearable devices add another dimension. A study integrating 24-hour wrist-worn activity data from UK Biobank participants with demographic information found that adding the raw activity features boosted a machine learning classifier’s ability to predict cognitive performance, raising the area under the ROC curve from 0.66 to 0.76.²⁰ Results like these hint at a future where passively collected data, the kind your smartwatch already generates, could feed into health-risk models at the population level.

Rare Diseases Need Biobanks Most

Rare diseases collectively affect hundreds of millions of people worldwide, but each individual condition may have only a few thousand patients in a given country. This creates a brutal research problem: not enough samples to study. Biobanks that deliberately collect from rare-disease populations help overcome that scarcity. China’s Peking Union Medical College Hospital rare-disease biobank, launched in 2016, now integrates specimens from over 440 institutions through national networks, using standardized protocols and ISO accreditation to ensure consistency.²¹ Italy’s Telethon Network of Genetic Biobanks has played a similar role, giving researchers access to meaningful sample sizes for conditions that any single hospital would see only rarely.²²

Without this kind of coordinated infrastructure, studies of rare diseases stay small, underpowered, and fragmented. Many orphan drugs, medications developed for rare conditions, owe their existence to the ability to pool samples across sites and countries.

Ethics, Consent, and the Privacy Question

When you donate a sample to a biobank, you are contributing material that might be used for research you cannot foresee at the time of donation. This raises a genuine tension in research ethics. Many biobanks have operated under a “broad consent” model: you agree that your sample can be used for a wide range of future studies, subject to ethics-board oversight. An alternative called “dynamic consent” uses digital platforms to keep donors informed and let them make choices about each new study as it arises.²³ Dynamic consent respects donor autonomy more fully but creates logistical challenges for biobanks managing millions of samples.

Privacy is the other major concern. Genomic data is inherently identifying: your DNA sequence is unique to you. Could someone re-identify an “anonymous” biobank participant by predicting their physical traits from their genome and matching those predictions to a known individual? A recent analysis tackled this question head-on, building a mathematical framework to estimate how realistic such an attack would be. The conclusion was more reassuring than the headlines might suggest. Under real-world conditions, the precision of re-identification by trait prediction was below 0.13%, and identifying carriers of sensitive genetic variants like APOE-ε4 (linked to Alzheimer’s risk) achieved useful accuracy only at extremely low recall.²⁴ The practical risk, in other words, is far lower than some earlier studies had claimed. That said, privacy protections need to evolve alongside re-identification technology, and biobanks typically layer technical safeguards with legal agreements and access controls.

The Diversity Gap

The most studied biobanks in the world are overwhelmingly European in ancestry composition. The UK Biobank, for all its power, enrolled participants who are predominantly white and British. This matters because genetic risk scores built on one population often perform poorly when applied to another. A disease-prediction tool trained on European-ancestry genomes may overestimate or underestimate risk for people of African, East Asian, or South Asian descent. Achieving diverse representation in biomedical data is not a nice-to-have; failure to do so actively perpetuates health disparities and can harm patients from underrepresented backgrounds when those tools are used clinically.²⁵

The All of Us program in the United States was designed with this problem in mind, deliberately oversampling communities historically underrepresented in biomedical research. Other efforts, like the pan-UK Biobank analysis mentioned earlier, are working to extract more ancestry-specific genetic signals from existing data. But the gap remains large, and closing it requires not just more diverse biobanks but also the trust of communities that have historically had reason to be wary of medical research.

Keeping the Lights On

Biobanks are expensive to build and even more expensive to maintain. Freezers consume electricity around the clock, liquid nitrogen must be replenished, staff must process and catalog incoming samples, and IT infrastructure for linked health records requires ongoing investment. Many biobanks were launched with initial grants but face uncertain long-term funding. The U.S. National Cancer Institute developed a Biobank Economic Modeling Tool to help facilities build cost profiles, set pricing for samples and services, and perform financial forecasting.²⁶

Some biobanks charge researchers a fee per sample to recover costs, while others are funded through national health budgets or public-private partnerships. The Spanish HIV HGM BioBank, for example, developed a detailed economic plan including cost-recovery targets by period and service type to sustain operations.²⁷ No single model has won out, and financial sustainability remains one of the field’s most persistent headaches. A biobank that runs out of funding does not just stop collecting; it risks losing the samples already stored, which can represent decades of irreplaceable participant contributions.

Cross-Border Sharing and Its Complications

Science benefits when biobanks share samples and data across national borders, but doing so is legally and ethically tangled. Different countries have different consent requirements, different rules about data protection (the EU’s GDPR being the most stringent), and different expectations around benefit-sharing with the communities that provided samples. A study of researchers involved in cross-border biobanking in Uganda found that contradictory legal frameworks between countries were a major barrier, and that many researchers negotiating material-transfer agreements did not fully understand the regulations on both sides.²⁸ The recommendation was straightforward: involve legal experts early and read the guidelines before negotiating. In practice, that kind of preparation is often skipped under the pressure to get research moving quickly.

Preserving Microbial Communities

Biobanking is expanding beyond human cells. The human microbiome, the trillions of bacteria, fungi, and viruses living in and on your body, is increasingly recognized as central to health. Preserving these complex communities for future research introduces challenges that differ from storing a blood sample. Microbial communities are diverse, with species that vary enormously in their sensitivity to freezing and thawing. A study on preserving fermented-food microbiomes found that long-term conservation of complex microbial communities remains difficult precisely because of this heterogeneity.²⁹ Some species survive cryopreservation well; others die off, skewing the composition of the thawed sample. Developing protocols that keep an entire microbial ecosystem intact, rather than just the hardiest members, is a frontier problem for the field.

Building and maintaining public trust is what makes all of this possible. Biobanks depend on voluntary donations, and people are more willing to participate when they understand what their samples will be used for, trust that their privacy is protected, and feel that the research serves their community. Strategies for building that trust range from community advisory boards and transparent governance to simply explaining biobank operations in plain language.³⁰ Without donors, none of the research described above happens. The freezers stay full only as long as people keep saying yes.