Velociraptor: Cross-Platform Quantitative Search Using Hallmark Cell Features.

Area of Science:

• Single-cell biology
• Computational biology
• Immunology

Background:

• Identifying novel cell types and validating them across studies and platforms is a major challenge in single-cell discovery analysis.
• Current methods struggle with automated validation due to reliance on hierarchical feature use and platform-specific rules.

Approach:

• Introduced Velociraptor, a cross-platform workflow using Marker Enrichment Modeling for cell identification, quantitative description, and searching.
• Velociraptor registers cells between datasets, including flow cytometry and quantitative imaging, using overlapping feature sets.
• The workflow identifies cells based on hallmark feature sets, not just per-feature cutoffs or hierarchical analyses.

Key Points:

• Velociraptor successfully identified disease-specific cells in glioblastoma and genotype-specific cells in inborn errors of immunity.
• Cross-platform analysis in breast cancer correctly identified memory T cell subsets using multiplex imaging.
• Phenotypic descriptions from algorithms or humans proved effective for search inputs, demonstrating flexibility.

Conclusions:

• Velociraptor enables robust cross-platform single-cell analysis, facilitating the discovery and characterization of novel cell populations.
• The workflow advances the understanding of cellular heterogeneity in disease states and immune responses.
• Velociraptor offers a flexible and powerful tool for single-cell discovery across diverse biological datasets and platforms.