Statescope: an integrative deconvolution framework for discovering cell states in tumors

Area of Science:

• Computational biology
• Cancer genomics
• Immunogenomics

Background:

• Accurate deconvolution of cell states from bulk tumor RNA sequencing (RNA-seq) is challenging due to malignant cell heterogeneity in cancer.
• Existing methods struggle to precisely identify distinct cell populations and their states within complex tumor microenvironments.

Purpose of the Study:

• To introduce Statescope, a novel Bayesian framework designed to enhance the accuracy of cell state deconvolution from bulk tumor RNA-seq data.
• To overcome limitations posed by malignant cell heterogeneity and inter-sample variation in cancer deconvolution.

Main Methods:

• Developed Statescope, a Bayesian framework integrating DNA-derived malignant cell purity.
• Incorporated explicit modeling of inter-sample variation to improve cell state identification.
• Benchmarked Statescope against existing deconvolution methods using comprehensive datasets.

Main Results:

• Statescope demonstrated superior performance in both cell fraction and state estimation compared to current methods.
• The framework successfully identified cell states not present in single-cell references, including novel neutrophil states in lung cancer.
• Statescope identified a combinatorial signature of effector CD8+ T cells and conventional dendritic cells predicting immunotherapy survival in clinical trials.

Conclusions:

• Statescope offers a robust solution for accurate cell state deconvolution from bulk RNA-seq, addressing critical challenges in cancer research.
• The framework advances deconvolution beyond mere estimation, transforming it into a powerful discovery platform for biological and clinical insights.
• Statescope enables deeper understanding of tumor microenvironments and immune responses, leveraging widely available multi-omics data.