The field in 2015
When GEN first covered our work in single-cell functional assays in 2015,1 immunology was entering a new era of molecular resolution. Researchers were gaining deeper insight into immune-cell heterogeneity through rapidly advancing single-cell technologies, reshaping how disease biology could be studied.
Tania Konry, PhDCo-founder, Feromics
Yet an important challenge remained unresolved: molecular measurements could describe what immune cells appeared to be, but they did not always predict what immune cells would actually do over time.
Immune cells with activated molecular profiles sometimes failed to sustain durable antitumor activity, while cells that expanded efficiently during manufacturing did not necessarily maintain persistence or therapeutic efficacy. The biological processes that determine therapeutic outcome unfold dynamically through behavior, adaptation, persistence, and dysfunction over time, requiring approaches capable of observing immune behavior directly alongside molecular state.
At Northeastern University, our laboratory focused on developing technologies designed to study immune function in a more dynamic and biologically relevant way. Much of this work centered on creating single-cell systems capable of observing immune-cell behavior in real time while preserving the biological context required for meaningful downstream molecular analysis.
Over the following decade, advances in microfluidics, live-cell imaging, computational analysis, and single-cell biology made it possible to connect directly observed immune behavior with downstream molecular and clinical information at scales that were previously unattainable. What began as an effort to observe immune-cell behavior more faithfully evolved into integrated functional-immunomics systems capable of supporting translational analysis, therapeutic characterization, and more sophisticated computational modeling.
These efforts contributed to the emergence of functional immunomics—an approach centered on understanding immune behavior and its relationship to disease and therapeutic outcome.
A shift toward functional immunomics
One of the major transitions in immunology over the past decade has been the recognition that immune-cell behavior represents a critical biological variable.
Traditional molecular approaches continue to provide essential insight into gene expression, signaling state, and cellular composition. Functional approaches extend those modalities, adding behavioral context to molecular information.
Persistence, resistance to dysfunction, therapeutic durability, and effective immune response are fundamentally functional properties. Behaviors such as serial killing, sustained cytotoxicity, and resistance to exhaustion cannot be fully captured through static measurements alone because they emerge over time.
Advances in microfluidics, live-cell imaging, and single-cell analysis made it increasingly possible to observe these processes directly. In our own work, controlled single-cell pairing systems enabled immune cells to be placed into defined microenvironments with tumor targets, allowing functional behavior to be observed before downstream molecular analysis was performed.2,3
This sequence—observing what a cell does before analyzing what it contains—represented an important conceptual shift. Rather than inferring function from molecular correlation alone, functional approaches make it possible to anchor molecular interpretation to directly observed biological behavior.4
In our early single-cell experiments, what struck me most was how often immune cells with strong activation signatures failed to sustain functional killing over time, and how often cells that appeared less remarkable molecularly turned out to be the ones driving effective cytotoxic responses. After seeing that pattern repeatedly across experiments, it became increasingly difficult to think about immune behavior as something a molecular state alone could fully explain. That observation gradually shifted my own perspective from focusing primarily on what immune cells contained to focusing on what they actually did.
A growing number of academic and industry groups are now contributing to this growing emphasis on function-linked immune analysis. Over time, our own work evolved into what we described internally as “Function-to-Omics”—a conceptual framework in which immune behavior serves as a foundational reference point for understanding disease and therapeutic response.
Much of the foundational intellectual property underlying this work originated through my research at Northeastern University and was later advanced translationally through Feromics, a functional immunomics company focused on AI-enabled immune analysis. The broader goal has been to establish function as a key framework for immune analysis, therapeutic development, and predictive modeling.
The AI inflection point
The rise of AI in biology has amplified the importance of this transition.
Machine-learning systems are fundamentally shaped by the biological quality and structure of the data used to train them. Bulk population-averaged datasets introduce biological noise by averaging across heterogeneous cell states, whereas function-linked analyses resolve this heterogeneity and yield cleaner, more therapeutically informative immune signatures.
A mixture of cells in different states—some highly cytotoxic, some exhausted, some transitional—may produce a population-level signal that reflects none of those states precisely. In practice, this means that cells expressing canonical activation markers may still fail to demonstrate sustained cytotoxic activity when observed functionally over time. Functional analysis makes it possible to separate those behaviors rather than averaging them into a single composite measurement.
As functional single-cell datasets mature, they create opportunities for computational systems capable of linking immune behavior more directly to therapeutic outcome.
More importantly, they change the structure of the learning problem itself.
Rather than training models primarily on correlative molecular associations, function-linked datasets allow computational systems to learn from experimentally observed biological outcomes at the level of individual cells.
This represents an important conceptual shift for the field: moving from systems built primarily on correlation toward approaches informed by experimentally observed biology.
At Feromics, this convergence between functional biology and AI is now being advanced through the development of large functional immunomics datasets linking immune behavior with downstream molecular and clinical information.5 Supported in part through a contract with the Advanced Research Projects Agency for Health (ARPA-H)6, these efforts have shown potential to predict response to immune-based therapies using function-labeled immune datasets linked to clinical outcome.
In these systems, models are trained on datasets in which observed immune-cell behavior—including cytotoxic activity, persistence, and resistance to dysfunction—is connected directly to downstream molecular and clinical data. The goal is to generate biologically meaningful datasets capable of supporting more predictive and clinically relevant computational models.
The long-term implication is the possibility of improved biological analysis and computational systems that better represent patient-specific immune behavior and therapeutic response.
Clinical translation
The clinical implications of functional immunomics are beginning to emerge across multiple areas of immunotherapy and translational medicine.
In cell therapy, conventional metrics alone often do not fully capture therapeutic potential. Functional approaches create opportunities to improve therapy characterization, donor evaluation, patient stratification, and response prediction through a deeper understanding of immune behavior.
This is already becoming visible in donor-derived immune-cell studies, including work in acute myeloid leukemia (AML) and lymphoma, where functional stratification prior to molecular analysis has begun to reveal distinct transcriptomic programs associated with antitumor activity. These types of approaches create opportunities to identify high-performing immune subsets based on directly observed function rather than molecular inference alone.
Therapy-response prediction represents another important area of development. Profiling patient immune cells based on how they behave in the presence of disease-relevant targets creates the possibility of predicting therapeutic response before treatment begins.
Functional approaches may also influence target discovery and biomarker development by connecting biological activity more directly with therapeutic outcome. In this context, molecular information becomes most powerful when interpreted alongside observed behavior rather than independently from it.
Several of these concepts are now transitioning from academic research into translational development. Through Feromics, technologies originating from my research at Northeastern University are being advanced toward applications in immunotherapy development, precision immune profiling, AI-enabled therapeutic prediction, and next-generation engineered immune-cell therapies.
Over time, functional immunomics may support more individualized therapeutic strategies and more predictive models of immune response across oncology, autoimmune disease, and broader immune-mediated disorders.
Looking ahead
The next decade will likely focus less on proving the value of functional biology and more on integrating it into scalable research and clinical frameworks.
Reproducibility, standardization, data integration, and clinical translation remain major challenges for the field. Functional biology is inherently complex and capturing immune-cell behavior in ways that faithfully reflect human disease remains an ongoing scientific challenge.
For much of modern immunology, the central challenge was describing the immune state with increasing molecular precision. The next phase of the field may be defined not only by molecular description, but also by the ability to measure how immune systems behave dynamically over time—functionally, adaptively, and in ways directly connected to therapeutic outcome.
The question that motivated much of our work in 2015—what is the immune system actually doing?—is now becoming possible to answer at single-cell resolution, at a meaningful scale, and in ways increasingly connected to clinical decision-making.
The convergence of functional biology, AI, and translational medicine is beginning to reshape how immune systems are studied, modeled, and therapeutically engineered.
References
Marusina K. Single Cell Is No Longer a Limit. Genetic Engineering & Biotechnology News. March 15, 2015;35(6).
Sharkey C, Akligoh H, Finocchiaro M, et al. High-throughput 3D matrigel-based droplet microfluidics for single-cell function-to-omics analysis of cytotoxic immune cells in solid tumor interactions. Adv Mater Technol. 2026;202501801.
Sullivan MR, White RP, Dashnamoorthy Ravi, et al. Characterizing influence of rCHOP treatment on diffuse large B-cell lymphoma microenvironment through in vitro microfluidic spheroid model. Cell Death Dis. 2024;15(1):18.
Sharkey C, White R, Finocchiaro M, Thomas J, Estevam J, Konry T. Advancing point-of-care applications with droplet microfluidics: from single-cell to multicellular analysis. Annu Rev Biomed Eng. 2024;26(1):119-139.
White R, Sharkey C, Vyas J, et al. Transcriptomic profiling of T cell immune states before and after donor lymphocyte infusion in a patient with acute myeloid leukemia. Blood. 2025;146(Suppl 1):2566.
ARPA-H Contract No. 75N91024C00036. Advanced Research Projects Agency for Health. (Contract awarded to Feromics Inc., $4.1 million; publicly disclosed.)
Tania Konry, PhD, is the cofounder of Feromics and associate professor at Northeastern University.
The post A Decade of Advances in Single-Cell Functional Immunomics appeared first on GEN – Genetic Engineering and Biotechnology News.
