Drug discovery has long been a high-risk, high-reward endeavor. Developing a new therapeutic from initial concept to market approval can take over a decade and cost more than $2 billion on average. Despite these staggering investments, the failure rate remains high, with many drugs faltering in late-stage clinical trials due to unforeseen toxicity, lack of efficacy, or poor patient stratification. Traditional approaches often focus on targeting a single molecule or pathway, an approach increasingly recognized as insufficient for tackling complex diseases such as cancer, neurodegeneration, and autoimmune disorders.
In this high-stakes environment, multi-omics technologies, combined with AI and machine learning, are proving to be powerful tools for accelerating and de-risking the drug discovery pipeline. Multi-omics refers to the integration of high-throughput molecular profiling technologies — including genomics, transcriptomics, proteomics, metabolomics, epigenomics, and more — to provide a holistic understanding of biological systems. Rather than viewing cellular processes through the lens of a single molecular type, multi-omics allows researchers to capture the interconnected web of biological activity, revealing disease mechanisms and drug responses that would remain hidden using traditional approaches.
By combining multi-omics with AI, researchers can move beyond trial-and-error approaches, pinpointing targets more accurately and designing experiments more efficiently. This integration accelerates early-stage discovery while also reducing the risk of late-stage clinical failures, offering a pathway toward more precise, personalized, and effective therapeutics.
Why multi-omics matters in drug discovery
Integrating multiple omics layers can validate promising therapeutic targets and filter out false leads before they derail expensive clinical trials.
—Juozas Gordevičius, VUGENE
One of the biggest limitations of traditional drug discovery is that it often tells an incomplete story. Looking at a single molecular layer — a gene, a protein, or a metabolite — can miss the broader context of how disease actually unfolds in the body. Multi-omics addresses this gap by integrating multiple layers of molecular information, offering a systems-level view of cellular activity.
Genomics can reveal mutations or inherited risk factors, while transcriptomics shows which genes are being turned on or off. Proteomics and metabolomics capture how those changes translate into functional proteins and chemical reactions, and epigenomics sheds light on the regulatory mechanisms controlling gene activity. Alone, each dataset tells only part of the story; together, they can map the complex networks and pathways driving disease.
For instance, a gene mutation that seems innocuous in isolation may trigger a cascade of downstream effects when considered alongside changes in protein levels or metabolic activity. “A differentially expressed gene does not necessarily translate into a differentially abundant protein,” Juozas Gordevičius, Chief Technology Officer and founder of VUGENE, told DDN. “Integrating multiple omics layers can validate promising therapeutic targets and filter out false leads before they derail expensive clinical trials.”
From data to insight
Today, as the cost of generating multi-omics data continues to fall, the real challenge lies not in collecting the data but in analyzing and making sense of it.
—Juozas Gordevičius, VUGENE
However, collecting multi-omics data is just the beginning. The real challenge lies in making sense of it. Each dataset — whether genomic sequences, protein abundances, or metabolite profiles — is often noisy, incomplete, or generated under different experimental conditions. Bringing these diverse layers together requires more than statistical know-how; it demands a deep understanding of biology, computational methods, and how the two intersect.
“Labs can generate diverse datasets but understanding that data is slow and complex. There’s a gap between computational expertise and biomedical research needs,” Gordevičius said. “Today, as the cost of generating multi-omics data continues to fall, the real challenge lies not in collecting the data but in analyzing and making sense of it.”
This is where AI and machine learning come into play. Advanced algorithms can fill in missing data, filter out irrelevant signals, and identify meaningful patterns that might escape even seasoned researchers. By iteratively analyzing pathways and networks across multiple molecular layers, scientists can pinpoint the mechanisms driving disease and identify targets for intervention.
“A key part of our work is creating customized models directly from our clients’ data,” said Gordevičius. “These models can help answer questions like whether patients can be accurately diagnosed, how outcomes might be predicted, or what the minimum set of biomarkers is to make those predictions. Providing these actionable models has become a standard part of our analysis — giving researchers tools they can continue to use and build on for future studies.”
By combining multi-omics with AI, researchers can move from raw data to actionable hypotheses in days rather than months. “Spatial and single-cell data are extremely data-intensive, and analyzing them requires both significant computational resources and specialized expertise — it’s not something that can be done in a standard lab,” said Gordevičius. “We aim to make these analyses accessible to researchers across industry and academia, democratizing access to advanced AI models so they can drive drug discovery and uncover mechanisms of action. Without this, such capabilities remain limited to the largest pharmaceutical companies, slowing the pace of scientific progress."
Revealing druggable targets
Real-world applications are already demonstrating the power of this approach. In a recent study published in Critical Care, Michigan State University, Corewell Health, and VUGENE applied longitudinal multi-omics to map the systemic biological response in children following severe traumatic brain injury (TBI). By integrating transcriptomics and metabolomics over multiple time points, the team captured a detailed, dynamic view of the molecular changes that unfold in the days after injury.
The analysis revealed 345 transcripts forming distinct temporal signatures, highlighting 50 potential biomarkers capable of distinguishing the body’s immediate trauma response from its recovery phase. Among these, the S100A8/A9 complex emerged as a mechanistic driver of neuroinflammation and oxidative stress. Notably, S100A8/A9 inhibition in preclinical models reduces neuroinflammation and neuronal death, though its therapeutic potential in pediatric and adult TBI remains largely unexplored.
Their integrated model also noted disruptions in the polyamine pathway, with increased expression of the ornithine decarboxylase gene correlating with elevated plasma putrescine levels. In rodent models of TBI, these changes have been associated with heightened inflammation, neuronal dysfunction, and vasogenic edema. Additionally, disruption of polyamine homeostasis is hypothesized to lead to a neurotoxic environment contributing to secondary brain injury, suggesting that strategies aimed at modulating polyamine metabolism could offer a promising therapeutic avenue.
The study also revealed systemic lipid surges and shifts in amino acid metabolism, reflecting the brain’s attempts to repair membranes and manage acute energy deficits. Notably, levels of indole-3-propionic acid, a metabolite produced by the gut microbiome, decreased in proportion to clinical severity, pointing to a potential role for the gut-brain axis in trauma recovery.
By integrating diverse datasets and applying computational modeling, researchers can identify potential biomarkers and therapeutic targets while also enabling the reconstruction of the dynamic pathways driving disease and recovery.
Future progress
As sequencing technologies and molecular profiling tools become faster and more affordable, multi-omics is poised to reshape the landscape of drug discovery. By offering a systems-level view of disease biology, these approaches enable researchers to move beyond the guesswork of traditional methods and make data-driven decisions about which targets to pursue, which patient populations to focus on, and which therapies are most likely to succeed.
The integration of AI and machine learning is central to this transformation. Algorithms can sift through vast datasets, highlight meaningful patterns, and even predict how a drug might behave in a complex biological system. This combination of multi-omics and computational power allows for a more predictive, rather than reactive, approach to drug development — one that has the potential to shorten timelines, reduce costs, and improve success rates in clinical trials.
