Researchers have developed a powerful new tool that makes it easier to study the mix of cell types in human tissue, which is crucial for understanding diseases such as cancer.
Developed by researchers at Oregon Health & Science University’s Knight Cancer Institute, the tool, dubbed OmicsTweezer, uses advanced machine learning techniques to analyze biological data at a scale large enough to estimate the composition of cell types in a sample of tissue that may be taken from a biopsy. This process allows scientists to map the cellular makeup of tumors and surrounding tissues – an area known as the tumor microenvironment.
They published their findings today in Cell Genomics.
“The tumor microenvironment, made up of diverse cell types that shape tumor development and patient outcomes, has been a longstanding research priority at the Knight Cancer Institute,” said senior author Zheng Xia, Ph.D., associate professor of biomedical engineering in the OHSU School of Medicine and a member of the OHSU Knight Cancer Institute.
“Our goal is to infer cell type composition using bulk data from large clinical sample sizes.”
Usually, scientists use data from the whole tissue (called “bulk data”) and try to compare it with data from individual cells to estimate the composition of cell types. But these two types of data often don’t match because they are collected in different ways. This mismatch, called a “batch effect,” can make it hard to get accurate results.
OmicsTweezer compares known patterns from single-cell data – where researchers can study one cell at a time – with the more complex, mixed data from bulk samples. It does this by aligning both types of data in a shared digital space, making it easier to match patterns and reduce errors caused by differences in how the data was collected, leading to more reliable results.
Overcoming limits of single-cell data
While single-cell technologies can provide detailed views of individual cells, they remain expensive and technically difficult to apply to large numbers of cells within tissue samples from patients. As a result, scientists often rely on more accessible bulk data, which averages signals from many cells.
It’s still very expensive to profile a large clinical sample size using single-cell technology. But there is an abundance of bulk data – and by integrating single-cell and bulk data together, we can build a much clearer picture.”
Zheng Xia, Ph.D., associate professor of biomedical engineering, OHSU School of Medicine
Traditional tools use a simpler linear model to estimate cell types based on gene expression. But OmicsTweezer takes a more sophisticated approach, using deep learning – a branch of machine learning that finds non-linear patterns in complex data – and a method called optimal transport to align different types of data.
“We use optimal transport to align two different distributions – single-cell and bulk data – in the same space,” Xia said. “In this way, we can reduce the batch effect, which has long been a challenge when working with data from different sources.”
New possibilities in cancer research
Researchers tested OmicsTweezer on both simulated datasets and real tissue samples from patients with prostate and colon cancer. It successfully identified subtle cell subtypes and estimated cell population changes between patient groups, which could help scientists pinpoint potential therapeutic targets.
“With this tool, we can now estimate the fractions of those populations defined by single-cell data in bulk data from patient groups,” Xia said. “That could help us understand which cell populations are changing during disease progression and guide treatment decisions.”
OmicsTweezer was developed as part of a multidisciplinary collaboration at the OHSU Knight Cancer Institute, in partnership with Lisa Coussens, Ph.D., FAACR, FAIO, Gordon Mills, M.D., Ph.D., and the SMMART project. SMMART stands for Serial Measurements of Molecular and Architectural Responses to Therapy. It is the flagship project of the Knight Cancer Institute’s precision oncology program, which helps identify new treatments that last longer and improve the quality of life for patients with advanced cancer.
“This kind of work wouldn’t be possible without collaboration,” Xia said. “It really reflects the strength of the team at the Knight Cancer Institute.”
Source:
Oregon Health & Science University
Journal reference:
Yang, X., et al. (2025). OmicsTweezer: A distribution-independent cell deconvolution model for multi-omics Data. Cell Genomics. doi.org/10.1016/j.xgen.2025.100950.
