Overview

• OmicsTweezer was peer-reviewed and published in Cell Genomics in July 2025 after successful tests on simulated datasets and real prostate and colon cancer tissue samples.
• It uses deep learning with optimal transport to align single-cell and bulk omics data, mitigating batch effects and reducing the need for expensive single-cell assays.
• Validation studies demonstrated its ability to identify subtle cell subtypes and quantify shifts in tumor and surrounding tissue populations, supporting targeted therapeutic research.
• The model was developed by a multidisciplinary team at OHSU’s Knight Cancer Institute in collaboration with the SMMART project and experts Lisa Coussens and Gordon Mills.
• Designed for scalability, OmicsTweezer accommodates multiple bulk omics inputs—such as RNA sequencing, proteomics and spatial transcriptomics—to profile tumor microenvironments in large clinical cohorts.