Situation
A small, clinical-stage biotechnology company developing next-generation cancer immunotherapies sought to identify a rare tumor-associated cell population believed to play a critical role in metastasis. These cells were hypothesized to contribute to immune evasion and tumor spread. However, identifying this population across multiple single-cell RNA-sequencing (scRNA-seq) datasets posed significant challenges due to their extreme rarity and the low abundance of key marker genes.
Compounding the complexity, the markers of interest were not reliably detected in standard scRNA-seq outputs, necessitating the use of a gene expression imputation strategy. Implementing this approach required careful optimization and extensive validation to ensure biologically meaningful results and avoid introducing artifacts. The company’s overarching objective was to characterize this rare cell population across tumor samples, compare their transcriptional signatures, and evaluate their potential as druggable targets across cancer types.
As a small company with limited internal bioinformatics capacity and budget constraints, they explored several external partners. One of their executives, who had previously worked with Bridge Informatics (BI) earlier in his career, remembered our team for consistently delivering high-quality, dependable work. Based on that experience and the cost advantages of working with BI over hiring a full-time bioinformatician, they ultimately chose to re-engage with us.
Strategy
BI partnered closely with the Company to understand their broader R&D goals before proposing a tailored, project-based engagement. We collaboratively scoped a plan that included a clearly defined set of deliverables and a fixed number of hours, giving the Company full visibility into cost and timeline from the outset. Together, the team designed and executed a robust, multi-step analysis specifically tailored to uncover and characterize the rare cell population of interest:
Curated Datasets and Conducted Quality Control
- Sourced matched primary and metastatic colorectal cancer scRNA-seq datasets from public repositories
- Performed stringent quality control to exclude low-quality cells and doublets using thresholds for mitochondrial reads, ribosomal content, and cell complexity
Identified Cells via Iterative Marker-Based Strategy with Imputed Expression
- Developed and refined a rigorous method for identifying the target cell population using a combination of raw and imputed gene expression data
- Applied gene expression imputation to address dropout noise common in single-cell data, enabling detection of cells with expression patterns that may otherwise be missed
- Employed multiple rounds of labeling and refined selection criteria across datasets using both canonical and custom gene panels
- Compared imputation-based annotations to those from raw expression data and confirmed high concordance through overlap analysis, increasing confidence in cell identification
Conducted Differential Expression and Pathway Analysis
- Conducted differential gene expression analysis between rare cells associated with primary tumors and those from metastatic lesions
- Identified unique gene signatures enriched in the metastatic context, including pathways related to immune modulation, cell migration, and antigen presentation
- Used multiple databases including KEGG, Reactome, GO, and WikiPathways to interpret biological significance and assess targetability
Conducted Functional Characterization
- Investigated ligand-receptor communication using CellChat to determine how the rare cell population interacts with its microenvironment
- Applied trajectory inference tools to model state transitions suggestive of epithelial-mesenchymal dynamics
- Examined pathway-level distinctions that could underlie functional differences between primary- and metastasis-associated cells
Presented Final Results with Visualizations
- Delivered CSV files containing ranked differentially expressed gene lists and pathway enrichment results
- Generated UMAPs, heatmaps, dot plots, and violin plots stratified by tissue region and annotation label
- Produced summary tables highlighting reproducible markers and shared pathways across datasets
Results
Within six weeks, BI enabled the client to clarify the molecular identity and metastatic potential of a rare, immunologically active tumor cell population:
Validated Cell Annotation Strategy
Achieved >90% overlap between cells identified using imputed vs. raw expression-based criteria. Delivered a reproducible framework for cell labeling across disparate scRNA-seq datasets.
Metastasis-Associated Gene Signature
Identified a cohort of genes consistently enriched in rare metastatic cells. Spotlighted several targetable pathways for further drug development efforts.
Pan-Cancer Implications
Demonstrated that several gene signatures were preserved across tumor regions and cell states. Proposed a strategy for expanding the analysis into other cancers to identify cross-indication therapeutic opportunities.
Conclusion
By combining rigorous data curation, gene imputation, and iterative labeling strategies, BI delivered a high-confidence transcriptomic profile of a rare cell population involved in metastasis, empowering the client to refine their drug development strategy and prioritize next-step validations for novel cancer targets.
This successful collaboration laid the foundation for an ongoing partnership, with the company returning for additional projects as new questions emerged across their discovery pipeline.
If your team is exploring complex biological questions and could benefit from tailored computational support, we welcome you to connect with us. Schedule a call to learn how Bridge Informatics can help advance your research.
