CACTUS

CACTUS maps genotype-to-phenotype relationships in tumors by integrating clonal architecture, genomic clustering, and single-cell transcriptome profiling to assign individual tumor cells to their clones of origin.

Key Features:

• Probabilistic Modeling: CACTUS employs a probabilistic model to enhance confidence in cell-to-clone assignments by leveraging multiple independent genomic measurements.
• Integration of Multiple Data Types: The tool integrates whole exome sequencing, single-cell RNA sequencing (scRNA-seq), and B cell receptor sequencing data for combined analysis.
• Clonal Architecture Mapping: CACTUS integrates clonal architecture with genomic clustering and transcriptome profiling at the single-cell level to map cells back to their clones of origin.
• Independent Genomic Clustering: The method uses an independent genomic clustering approach to define genomic clusters used for assignment.
• Sparse scRNA-seq Support: CACTUS operates with sparse scRNA-seq data to enable mapping despite limited transcript coverage per cell.
• Application to Follicular Lymphoma: The approach is applied to follicular lymphoma, leveraging mutation patterns in exome and B cell receptor loci to study clonal evolution.
• Improved Performance Over Predecessors: Incorporation of additional genomic data sources improves assignment accuracy of cells and B cell receptor–based clusters to tumor clones compared to previous models.
• Enhanced Model Certainty: Integration of diverse measurements increases the overall certainty of clone-assignment predictions.

Scientific Applications:

• Understanding Tumor Heterogeneity: CACTUS facilitates detailed mapping of genotype-to-phenotype relationships at single-cell resolution to study functional consequences of tumor heterogeneity.
• Origins of Resistance to Therapies: By charting clonal evolution and assigning cells to clones, CACTUS aids in identifying mechanisms and origins of resistance to targeted therapies at the single-cell level.

Methodology:

CACTUS applies a probabilistic model together with an independent genomic clustering approach combined with sparse scRNA-seq data to map single cells to imperfect genotypes of tumor clones.