Single-cell sequencing and imaging technologies are ever-improving. However, some challenges still stand in the way of meaningful, actionable insights. The Cellanome R3200, using Cell CageTM technology, offers a solution to:
Difficulty matching imaging data with single-cell sequencing results
Inability to identify active or quiescent cells without extensive bioinformatics
Limited ability to pinpoint resistant cell populations and the transcripts driving them
Challenges linking differentiated cell states to their corresponding transcriptomic profiles
Transcriptomic hits that don't validate functionally
Marker proxies that don't capture the full functional phenotype you're interested in
In this article, we break down how Cellanome achieves these results, the data it collects, and examples of the platform in use.
Fig. 1. Integrated experimental and computational workflow linking morphology and transcriptomics.Schematic of Cellanome’s CellCage™ technology, which enables multi-modal phenotypic and functional profiling of the same live cells. Tens of thousands of cells are mixed with a hydrogel precursor and loaded into a multi-lane flow cell. Individual cells are located, and CellCage™ enclosures (CCEs) are automatically generated around them via light-guided polymerization. The CCEs are permeable to small molecules and antibodies, allowing long-term culture and time-course imaging of the same cells.
Fig. 2. Schematic of 8-lane flow cell. Tens of thousands of cells are mixed with a hydrogel precursor and loaded into a multi-lane flow cell.Adherent or suspension cells are loaded into a 4-8 lane flow cells coated with the appropriate growth matrix and oligo dT for mRNA capture.
CellCageTM Enclosures (CCEs) are micro-3D-printed “cages” that isolate cells or ensembles of interest. The Cellanome platform uses AI-powered workflows to generate CCEs around individual single cells. CCEs are permeable to drugs, stains, and antibodies and images can be taken at multiple timepoints during a treatment, activation, or differentiation.
Fig. 3. Individual cells are located, and CellCage™ enclosures (CCEs) are automatically generated around them via light-guided polymerization. The CCEs are permeable to small molecules and antibodies, allowing long-term culture and time-course imaging of the same cells.
At the endpoint, cells are lysed within their CellCages, and reverse transcription (RT) is used to generate barcoded cDNA within each single cell’s enclosure. Following RT, CCEs are washed away and cDNAs are pooled for library preparation and sequencing.
Image processing and feature extraction: Analyze all images to extract fluorescence intensity and morphology in all objects. These results, along with cropped images of each CCE, are automatically uploaded to Cellanome Cloud.
Gating: On Cellanome Cloud, using fluorescence intensity and morphology, define object-level populations by “gating” data (i.e., manually selecting objects that have characteristics of interest). Compute CCE-level statistics, including counts of each cell population and average fluorescence of each cell population.
Multi-modal data integration: Cellanome Cloud automatically integrates multi-modal data at the CellCage™ structure-level into a single data structure. This enables you to link images, cellular features, and processed sequencing data measured in tens of thousands of cells over multiple time points.
Cellanome technology has been applied to several groundbreaking research projects, in many cases uncovering biological insights undiscoverable by other platforms. In the examples below, we walk through some of the most exciting discoveries powered by Cellanome CellCagesTM.
Hetereogeneity remains a major challenge in predicting the efficacy of cancer treatments. In this example, A549 cells (lung adenocarcinoma) were plated in the Cellanome.
CCEs were used to isolate individual single cells. Cells were then treated with 10µM of the EGFR inhibitor Olmutinib, and imaged longitudinally for 32 hours. Four unique imaging-based phenotypes were identified, including daughter cells that showed resistance to Olmutinib. All caged cells were subjected to single-cell RNA sequencing.
The linked imaging and sequencing data together were used to identify 15 unique pathways involved in EGFR resistance that would not have been identified by either modality alone.
Fig. 4. UMAP visualization and unsupervised clustering of the transcriptomic data.
Fig. 5. UMAP visualization and unsupervised clustering of the transcriptomic data colored by the phenotypic classification.
Fig. 6. Dot plot showing relative expression of unique targets associated with resistance. The Daughter Cell Resistant class is over-represented in cluster 5 (Fig.4), which includes p53 transcriptional targets associated with a quiescent/slow-cycling drug resistant state.
In this example, scientists use longitudinal imaging combined with transcriptome sequencing to monitor differentiation of preadipocytes to mature adipocytes producing lipid droplets. Images were collected before initiation of differentiation, and at multiple timepoints during differentiation, maintenance, and at full maturation. Staining with BODIPY (fluorescence staining of neutral lipids) is used to monitor preadipocyte differentiation to maturity.
Fig. 7. Overview of preadipocyte differentiation process and experimental workflow.
Single-cell transcriptomes were collected, but differential gene expression did not correspond to different stages. When BODIPY-stained cultures were analyzed to the UMAP, concordance of transcription was overall absent.
Fig. 8. BODIPY-stained cultures at various stages of differentiation indicated on UMAP .
Transcriptomic data alone did not match differentiation phenotypes detected by imaging (Fig. 8). An elastic net regression model predicting lipid accumulation from gene expression identified predictive genes for lipid accumulation that sequencing alone failed to identify. Less than 10% of the genes identified by this model overlapped with the top-20 differentially expressed genes between transcriptomic clusters, demonstrating that the genes driving functional phenotypes are largely distinct from those defining transcriptomic identity.
The model’s top three positive predictive target genes are uniformly distributed across the transcriptome (Fig. 9).
The above case studies are merely the tip of the iceberg for novel experiments that can be conducted using the Cellanome R3200. This collection of posters encompasses a broad range of assays and experimental designs, and we encourage you to check out these groundbreaking studies.
Cellanome has been tested with adherent, suspension, and co-cultures and produced excellent results. Below are some of the cell types that have been used routinely.
| Category | Cell Types |
| Immune Cells |
|
| Cancer Cells |
|
| Epithelial & Endothelial |
|
| Mesenchymal & Connective Tissue |
|
| Neuronal & Glial |
|
| Plants |
|
Psomagen is a trusted CRO offering:
Over 20 years of experience
CAP accreditation and CLIA certification
Deep scientific expertise in all aspects of multiomics
Cutting-edge technologies
Contact Psomagen today to learn how we can support your Cellanome research project. And between April 20 and June 30, 2026, enroll in our subsidized pilot program.
Images used with permission of Cellanome.