Can a tree tell us about a rainforest’s ecology, or a brick about a building’s architecture? Given just a few puzzle pieces, can we clearly grasp the full picture they belong to?
Biomedical researchers face that struggle with the human body, given its trillions of cells and hundreds of thousands of protein variants, directed by instructions from around 20,000 unique genes. For decades, technological limitations meant scientists could only investigate a small fraction of these components to glimpse their complex relationships with each other and with health and disease.
Enter the multi-omics revolution: an emerging class of cutting-edge technologies enabling unprecedented holistic views of biology. These tools not only survey whole biological landscapes—every gene in a genome, every protein in a proteome—but also reveal hidden links between them. From high-throughput gene sequencing technologies to artificial intelligence (AI) models that harmonise different data modalities, multi-omics approaches are helping unearth fundamental biological insights and advance effective, tailored healthcare interventions.
“By integrating high-quality molecular data, multi-omics research provides deep insights into pre-disease and disease mechanisms, as well as population-specific biomarkers,” said Lisa Ooi, Assistant Chief Executive of A*STAR’s Biomedical Research Council (A*STAR BMRC).
Several A*STAR research groups are spearheading multi-omics initiatives with a focus on Asian populations to fill significant gaps in regional representation in international studies. Through active, cross-border collaborations with academia, industry and government, A*STAR teams are also contributing to Singapore’s national precision health agenda and nurturing the growth of a regional multi-omics R&D hub.
Tools of the trade
Once extremely niche tools, omics profiling technologies are a cornerstone of many research institutions today. Whether genes, RNA transcripts, proteins or metabolites, the ability to sequence entire ‘-omes’ has led to an explosive emergence of publicly-available datasets.
These treasure troves of omics data have likewise driven analytical innovations, as teasing out true biological signals—and therefore clinically-meaningful insights—amid the noise can prove difficult depending on data quality and quantity. Among these innovations are AI models, which are increasingly indispensable parts of omics analytical pipelines.
“Instead of looking at one measure at a time, AI systems can blend many layers of biological information to paint a fuller picture of health and disease, such as identifying which gene-protein combinations might signal early-stage cancer,” said Dennis Wang, a Senior Principal Investigator at the A*STAR Institute for Human Development and Potential (A*STAR IHDP) and A*STAR Bioinformatics Institute (A*STAR BII).
These computational approaches are also helping uncover once-hidden insights from existing datasets. Researchers led by Jayantha Gunaratne, a Senior Principal Investigator and Deputy Division Director (Cell and Molecular Therapy) at the A*STAR Institute of Molecular and Cell Biology (A*STAR IMCB), developed a stepwise machine learning (ML)-based feature extraction pipeline called Unique Marker-AI (UMAI) to re-analyse published proteomics data. Through UMAI, they identified serum biomarkers of high-grade serous ovarian cancer, highlighting the translational value of AI-powered analytics in early disease detection.
“By integrating multiple validated algorithms, the pipeline effectively reduced analytical noise and prioritised biologically-relevant features. This led to the development of a novel four-biomarker panel that, when combined with current clinical markers, significantly enhanced diagnostic performance,” said Gunaratne.
Such omics-powered workflows are already addressing unmet diagnostic needs through strong collaborations with local and international clinical partners. Gunaratne’s clinical proteomics pipeline has uncovered biomarkers for several diseases, including a patent-pending urine biomarker panel for chronic kidney disease anchored on detecting earlier stages of kidney injury; prognostic serum markers predicting treatment response in chronic hepatitis B; and an immune-cold biomarker for breast cancer stratification. The group is also developing biomarker panels for a newfound high-risk subtype of breast and lung cancers.
Pinning locations
Advancements in omics throughput (or processing rate) and resolution now allow researchers to read entire genomes within hours. Bulk sequencing and single-cell omics provide a wealth of data about how individual cells, tissues and organs function in health and disease, as changes in gene expression profiles reflect the up- and down-regulation of biological pathways.
However, these methods can be limited by their samples, which are typically cells removed from their original environments. Spatial omics—the latest arrival in omics innovations—now adds proximity and location into the picture, mapping cellular gene expression profiles to their anatomical locations.
“Given much of a tissue’s function is encoded in how its cells interact, we need to know how its cells are arranged in living tissue,” said Shyam Prabhakar, Associate Director of the A*STAR Genome Institute of Singapore (A*STAR GIS). “Spatial omics shows us both the physical proximity and the mix of molecules within and around those cells, allowing us to infer cell-to-cell communications.”
With Singapore General Hospital (SGH)’s Tony Lim, the National Cancer Centre Singapore (NCCS)’s Iain Tan, A*STAR GIS’s Kok Hao Chen and A*STAR BII’s Hwee Kwan Lee, Prabhakar and colleagues spearheaded the establishment of high-resolution spatial omics for colorectal cancer research at A*STAR GIS through the SCISSOR programme. From 2020 to 2024, the team built a full spatial omics pipeline, running from tumour sample collection and preparation to sequencing and data analytics. This led to the inhouse development and refinement of tools such as mFISH, a molecular probing technique to simultaneously profile thousands of gene targets within a tissue’s spatial context; and BANKSY, an algorithm to analyse spatial omics data through integrated cell typing and tissue domain segmentation.
Through SCISSOR, the team discovered a rare invasive cell type that potentially drives tumour growth and metastasis in colorectal cancer. “As only a tiny population of this cell type exists in colorectal tumours, it was lost in the crowd of single-cell data. But spatial omics revealed that it forms a very thin, well-defined layer that pushes out from tumours into surrounding tissues,” said Prabhakar, adding that further findings in this area were pending publication.
Following SCISSOR’s success, A*STAR GIS, A*STAR BII, SGH and NCCS are jointly applying spatial omics to five cancers and two precancerous conditions through the TISHUMAP project, supported by a collaboration between A*STAR GIS and life science tech company 10x Genomics.
Across cancer types, profiles of a tumour’s surroundings, or microenvironment, can provide insights on cancer’s interactions with our immune system. Immune components such as natural killer (NK) cells play a critical role in disease progression, treatment response and even the risk of recurrence. For example, in hepatocellular carcinoma (HCC)—the most common form of liver cancer—NK cell infiltration into the tumour space is a key predictor of disease prognosis.
By combining multiple spatial omics methods, a team led by A*STAR IMCB Group Leader Joe Yeong and Senior Research Officer Denise Goh found a clinically-important subset of NK cells and five biomarkers whose spatial distribution patterns correlated with HCC recurrence. This led to the development of TIMES, a spatial immune scoring system that integrates spatial transcriptomics, spatial proteomics, multiplex immunohistochemistry and AI-driven analysis to generate a recurrence risk score based on the spatial expression of key biomarkers.
“TIMES enables the detection of molecular changes within specific cell types and tissue regions, allowing for more efficient identification of cancer biomarkers with biological relevance and spatial context,” said Yeong and Goh.
Choose your fighter (drug)
Just as the same disease can progress differently in different people, the same medication can elicit varying responses based on a whirlwind combination of genetics, metabolic processes, immune profiles and environmental factors. By shedding light on such diversity, multi-omics profiling stands as a critical enabler of both drug development and treatment selection within Singapore’s precision health agenda, starting with the issue of individual drug resistance.
“People tend to think resistance is the result of random mutations that change drug targets after starting treatment, but we’re starting to see resistance can be there before treatment begins,” said Sin Tiong Ong, a Professor at Duke-NUS Medical School’s Cancer and Stem Cell Biology (CSCB) Programme, as well as an A*STAR GIS Principal Investigator.
Ong explained that patients with chronic myeloid leukaemia (CML) carry the BCR::ABL1 driver gene mutation, creating an abnormal protein kinase that prompts the cancer’s aggressive growth. Tyrosine kinase inhibitors (TKI) are drugs designed to inhibit these proteins; however, actual treatment responses can vary from highly effective to dangerously ineffective between patients, with some leukaemias transforming into a treatment-resistant blast crisis stage.
To investigate this variability in CML patient response, Ong teamed up with Shyam Prabhakar and A*STAR GIS researchers, as well as SGH Senior Consultant Charles Chuah, Duke-NUS Principal Research Scientist Vaidehi Krishnan, and Immunoscape Associate Director Florian Schmidt. By combining single-cell RNA sequencing, flow cytometry and computational methods, they found that leukaemic stem cells—especially certain transcription factors within them that guided their differentiation trajectory—were significant predictors of TKI response. Patients who responded well tended to have stem cells destined to become red blood cells rather than white blood cells.
“That revealed a whole new way of thinking: the types of stem cells in which CML occurs could determine a patient’s sensitivity to TKIs,” Ong explained. “They can help us predict which patients are likely to undergo blast crisis months, even years ahead. We can then stratify these patients for closer monitoring and transplant them in earlier CML phases, potentially boosting survival rates.”
The team is now developing a flow cytometry kit to identify those at high risk of blast crisis transformation. Employing a cocktail of antibodies developed with A*STAR GIS data by the Duke-NUS CSCB team, including Research Fellow Mengge Yu, the kit is being validated in patients from Singapore and Australia, the latter in collaboration with Tim Hughes, Clinical Director of Precision Cancer Medicine at the South Australian Health and Medical Research Institute.
Given that evolving cell fates and functions are also critical to how patients react to drugs, Ying Swan Ho and researchers at the A*STAR Bioprocessing Technology Institute (A*STAR BTI) are harnessing a range of omics technologies to map metabolites: the chemical building blocks and messengers of cellular activity.
“Metabolic processes provide insights into how cells operate under normal conditions and how they change in response to disease and treatment, which allows us not only to discover drug targets, but ensure that new drugs are effective and safe,” said Ho, an A*STAR BTI Senior Principal Scientist.
Metabolomics can be critical when developing biotherapeutics. A class of products comprising antibodies, cells, RNA and other complex biological structures, biotherapeutics face ongoing challenges in production due to variability in quality and performance between manufacturing batches.
Ho believes that an expanded understanding of cell behaviour and advanced profiling techniques could make biotherapeutics production more consistent by identifying key quality indicators. For instance, metabolite data that reflects intracellular states and nutrient use can help producers optimise cell culture media to mimic bodily environments, promoting healthy and consistent cell growth.
To meet these needs, A*STAR BTI’s BioStream programme features a comprehensive suite of proteomics, metabolomics and glycomics capabilities to characterise biotherapeutics and the processes needed to manufacture them at scale. Supported by AI integration and mechanistic modelling in collaboration with Yang Zhang, Professor of Computer Science at the National University of Singapore, BioStream’s integrated approach is also complemented by A*STAR BTI’s proprietary display-and-secretion system for the widely-used CHO immortalised cell line, which enables high-throughput screening of complex biologic candidates as full molecules.
“This multi-dimensional analysis and preservation of molecular structural integrity helps us understand how structure, function and process conditions interact, enabling more informed development decisions,” Ho said.
Health across the lifespan
Beyond uncovering the mechanisms behind disease progression and treatment responses, multi-omics can be powerful trackers and predictors of health trajectories, supporting more preventive and proactive national healthcare models.
“Longitudinal multi-omics profiling platforms can reveal how factors like diet, lifestyle, microbiome and drug interventions interact to influence individual health, helping to identify disease risk even before symptoms emerge. These insights drive preventive health efforts that optimise health and promote healthy longevity,” Lisa Ooi said.
Ooi noted how research findings from the nationwide GUSTO birth cohort study have directly informed policymaking for gestational diabetes, a condition that poses both acute and long-term health risks for mother and child. National clinical guidelines were revised to now recommend universal screening during pregnancy, allowing for earlier detection and interventions against this glucose tolerance disorder.
“Beyond risk screening, GUSTO also uses omics data to uncover molecular drivers of maternal and child health,” Ooi added. “For example, linking fetal cord DNA methylation to maternal glucose levels of GUSTO mother-infant pairs revealed distinct epigenetic signatures associated with specific glycaemic traits, suggesting that different types of maternal hyperglycaemia can uniquely programme the neonatal epigenome and potentially influence early-life metabolic disease risk.”
Proteomics data from GUSTO also led Dennis Wang and A*STAR IHDP colleagues to discover the involvement of a protein called ephrin-A4 in early linguistic development, tracing its role to pathways that promoted myelination in brain regions for language-related functions. Myelination is a vital process not just during neurodevelopment but also in ageing, as it wraps lipid-rich sheaths around nerve fibres to enable rapid signal transmission and provide metabolic support to brain cells.
“Our findings are not just relevant for children; they may also help stratify dementia patients facing language difficulties,” Wang said.
To forecast evolving health conditions, Wang and colleagues are also building ML-based platforms such as GenMetS, a prediction model using genomic data for metabolic health status in young and healthy Asian adults. GenMetS has been validated in over 680,000 individuals from diverse cohorts including the UK Biobank, Japan Biobank, Chinese Kadoorie Biobank and Singapore’s ATTRaCT.
“We found that GenMetS can forecast a person’s risk of cardiometabolic disorders up to 30 years before clinical onset, and remains predictive across their lifespan, highlighting its potential for early intervention and lifelong health monitoring,” said Wang.
In brief: A*STAR multi-omics strategic partnerships
- Public healthcare clusters (SingHealth, NUHS, NHG)
- National programmes (STCC, CADENCE)
- National translation platforms (MedTech Catapult, DxD Hub, NATi, EDDC)
- Public-private partnerships (biopharmaceutical, life science tools and AI companies)
Driving clinical translation
Omics continues to reveal new depths to human diversity with powerful implications for precision healthcare. “What is ‘normal’ in a healthy population, when populations can be very different?” said Shyam Prabhakar.
Referring to the recently-published Asian Immune Diversity Atlas (AIDA), which found that immune cell compositions could vary as widely between Asian ethnic groups as between the sexes, Prabhakar highlighted the need to broaden clinical research coverage to more diverse and often underrepresented groups. Launched in 2019, AIDA’s initial collaborators included A*STAR GIS, RIKEN Japan and the Samsung Genome Institute.
Through projects such as AIDA, A*STAR research groups and partners across academia, industry and clinics continue to expand the foundation on the dynamic mechanics behind health and disease, advancing precision healthcare approaches for Singapore and beyond.
“Discoveries enabled by multi-omics can translate to risk-based screening and precision therapeutics, enabling earlier detection and targeting the right individuals at the right time with the right interventions,” Ooi said.
For Ooi, these tight clinical linkages and strong national translation platforms, further empowered by emerging AI capabilities, position A*STAR as a key innovation player in domestic and global multi-omics spaces. Whether through diagnostics, health monitoring or therapeutic innovations, A*STAR groups and collaborators jointly act as a translation engine to transform foundational omics research into improved clinical outcomes, simultaneously promoting scientific advances, bolstering the biotech scene and nurturing healthier communities.
