Omics and multiomics technologies are among the most dynamic fronts in biomedical research today: single cell omics emerged roughly a decade ago, while spatial omics is exploding right now. We spoke with Giovanni Tonon, full professor at Vita-Salute San Raffaele University (UniSR), where he also leads the Functional Genomics of Cancer laboratory. Omics and multiomics technologies make it possible to observe thousands of molecules of the same type simultaneously, in a single experiment, overcoming a constraint that shaped molecular biology for decades: the ability to study only one molecule at a time. Tonon has spent years running projects that combine sequencing, artificial intelligence and real clinical data. One of the goals: predicting how a tumor will respond to treatment before therapy even begins.
Until a few decades ago, molecular biology moved one unit at a time. «What we could do was study one molecule at a time, one gene at a time, one protein at a time», Tonon notes. At most, researchers could observe how two or three proteins related to each other, but no further.
Omics technologies broke through that barrier. «What omics does is see all the molecules of a given type simultaneously, or a large share of them», Tonon explains: not a single RNA molecule, but the entire RNA content of a cell at a given moment.
The plural, omics, reflects the fact that each molecule type requires its own technology: genomics for DNA, transcriptomics for RNA, proteomics for proteins, for example.
Multiomics goes a step further: instead of analyzing DNA, RNA and proteins separately, it observes them within the same experiment to understand how they relate to each other. It's a younger field. «Multiomics isn't very developed yet, we're only taking the first steps now, while the individual omics technologies are much more advanced» Tonon adds.
These are two distinct developments within the same field. The individual omics technologies, spatial omics included, are now mature and growing fast. Integrating all of them into a single analysis, true multiomics, remains a frontier that's still largely uncharted.
A concrete example of this emerging area is scGPT, an artificial intelligence model published in Nature Methods in 2024 and trained on more than 33 million single-cell profiles, designed specifically to integrate data from multiple omics types within the same model.
A tissue contains billions of different cells. Analyzing it "in bulk" produces averaged results that mask the differences between cell types: it would be like asking for the average composition of an eye, made up of dozens of distinct cell types, and getting back a single number.
Single cell omics, which emerged about ten years ago, solves this problem by observing gene expression cell by cell. «What single cell does is a huge step forward because it lets you analyze most of the genes,» Tonon explains, «even though it's not as sensitive as classic omics done on bulk tissue, meaning tissue analyzed as a whole, without distinguishing the cells that make it up».
This technical distinction reveals a real limit of the field: DNA and RNA are both nucleic acids and can be read with related sequencing technologies; proteins can't, because they're made of amino acids and require different methods, such as antibodies or mass spectrometry. «It's not that the technologies are less advanced: it's that protein biology is far more complex to study» Tonon observes. That's why a multiomics analysis combining RNA and DNA is easier to carry out today than one that also includes proteins.
The reach of these technologies is visible on an international scale too. In November 2024, the Human Cell Atlas consortium published a collection of more than 40 studies in Nature and related journals, the work of over 3,600 researchers across more than 100 countries, based on the analysis of roughly 100 million cells from more than 10,000 people, and built specifically on single cell and spatial omics. The first complete draft of the atlas is expected during 2026.
The latest frontier is spatial omics. Unlike single cell, which breaks tissue down into separate cells, spatial omics analyzes a tissue slice while keeping the position of every cell intact.
During the interview, Tonon reached for a comparison already common in science communication to distinguish the two techniques: a smoothie, which blends everything together, versus a fruit salad, where each piece stays where it is.
Today's resolution goes beyond the single cell. «With the new technologies you can see, for example, whether an RNA is more expressed in the nucleus than in the cytoplasm» Tonon notes, «a level of detail unthinkable with the omics technologies of ten years ago». Nature Methods named spatial transcriptomics "Method of the Year" in 2020, and spatial proteomics again in 2024, a double recognition that's rare for the same research area in the history of the award.
Tonon draws a line between two kinds of usefulness for omics and multiomics technologies: the scientific and the clinical. «There's no doubt, omics and multiomics technologies are fundamental, in the sense that they finally let us really see what's happening inside a cell: before, that wasn't possible» he observes.
Clinical translation, though, remains limited for now. In oncology, companies already sell tests based on gene expression panels, applied for example to breast cancer: applications that so far haven't had the hoped-for impact, the professor explains. The picture is different for personalized medicine based on identifying a handful of specific genes in a patient's tumor to guide therapy, and for whole-genome sequencing, increasingly common in prenatal diagnosis too.
Tonon is currently working on two projects that illustrate well the applications and potential of omics and multiomics. The first grows out of a collaboration with Professor Carlo Tacchetti, full professor of Human Anatomy at UniSR, who coordinates the S-RACE clinical artificial intelligence platform together with Professor Antonio Esposito. «We're trying to integrate clinical data with spatial transcriptomics: it's a path very few people are taking right now» Tonon says.
The goal is to more reliably predict how a disease will progress in an individual patient before treatment even starts. «The clinician's big problem is knowing, before it happens, whether it makes sense to treat a patient following a given protocol» explains the specialist: putting someone through surgery that's already known not to be curative carries a cost in quality of life without real benefit.
The project draws on both S-RACE and DIGICORE, the network of roughly 40 European oncology hospitals chaired by Tonon and set up to collect clinical data rigorously, producing what the literature calls Real World Evidence. That means observing how drugs actually work in the real population, not in clinical trials, where enrolled patients, the so-called "Olympic Patients," have a quality of life and physical performance far above average.
The second project is lab-based, carried out with the Politecnico di Milano on a microfluidics platform: a system of miniaturized chips that makes it possible to grow organoids and test multiple drugs in parallel on minimal amounts of tissue, automating steps that would otherwise take far more time and biological material in the lab.
The aim is to screen tumor organoids. For decades in oncology, it was nearly impossible to keep cells taken from a patient alive in culture: they died within a short time.
«Over the last ten, fifteen years there's been a huge conceptual shift: if you mix these cells, freshly taken from the patient, with a gel derived from murine tissue and with growth factors, you can grow them indefinitely» Tonon says.
These organoids have proven to be a valuable indicator: «They correlate very closely with patient survival: if the organoid responds to a given drug, the patient usually responds to that drug too».
A 2018 study in Science (Vlachogiannis et al.), for example, conducted on organoids derived from patients with metastatic gastrointestinal cancers, measured 100% sensitivity and 93% specificity in predicting, through the organoid, the patient's actual response to the same drug.
On these organoids, the laboratory led by Tonon applies a technique called Cell Painting. «It's essentially staining cells using eight different markers, which localize to different organelles» Tonon explains.
A drug treatment changes the morphology of these organelles in specific ways, and these morphological changes are in turn linked to transcriptional changes — the point of contact between Cell Painting and omics. The goal is «to screen patient-derived cells against drugs, measure them with Cell Painting and validate gene expression profiling, so we can identify an effective drug before therapy even starts.»
The technique, first described by the group at the Broad Institute of MIT and Harvard (Bray et al., Nature Protocols, 2016), is today one of the most widely used morphological profiling protocols in cancer research and drug screening.
«With the rise of artificial intelligence, a lot has changed» Tonon observes. «Tools like ChatGPT, now capable of writing code, have made the career path less linear than it was just a few years ago, when the road to becoming a bioinformatician was more predictable. It's a sign of a fast-moving field, where the skills required change faster than they used to».
One skill, though, is set to stay central: the ability to move between different professional languages. «I think integrating different worlds is an important path to success» he says: those who can speak with both a clinician and a bioinformatician, and get them talking to each other, have a real advantage.
That's exactly what the two projects described above put into practice. Integrating clinical data with spatial transcriptomics alongside Tacchetti means getting two groups that normally work with different tools and languages to collaborate; the same applies to the project with the Politecnico di Milano, where cell biology and materials engineering converge on the same platform.
«Understanding omics, and above all a quantitative approach to biology, is essential for a medical student today, and for a biotechnology student too» Tonon says.
That's the logic behind the course taught in English at UniSR's School of Medicine, called The Scientific Method, Quantitative Biology and the Art of Medicine, designed to integrate the quantitative and clinical sides of the profession from the earliest years of training.
Omics and multiomics technologies, in this sense, tell the story of a rapidly expanding field: laboratory tools keep changing, and with them the professional skills that anyone entering the field today will have the time and room to build.
Nature Methods, "Method of the Year", nature.com/nmeth
Vlachogiannis G. et al., "Patient-derived organoids model treatment response of metastatic gastrointestinal cancers", Science, 2018 (doi:10.1126/science.aao2774)
Bray M.A. et al., "Cell Painting, a high-content image-based assay for morphological profiling using multiplexed fluorescent dyes", Nature Protocols, 2016 (doi:10.1038/nprot.2016.105)
Human Cell Atlas Consortium, paper collection, Nature and Nature Portfolio journals, November 2024 (nature.com/collections/jccbbdahji)
Cui H., Wang C., Maan H. et al., "scGPT: toward building a foundation model for single-cell multi-omics using generative AI", Nature Methods, 2024 (doi:10.1038/s41592-024-02201-0)