ScSpTITH: a rank-correlation framework for robust quantification of multi-dimensional tumor heterogeneity

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Intra- and inter-tumoral heterogeneity (ITH) is a key driver of cancer progression and therapeutic resistance.
• Single-cell RNA sequencing (scRNA-seq) offers high resolution but suffers from significant "dropout" artifacts, limiting accuracy.
• Existing methods struggle to reliably quantify heterogeneity in complex cancer datasets.

Purpose of the Study:

• To develop a robust computational framework, ScSpTITH, for quantifying ITH in both scRNA-seq and spatial transcriptomic data.
• To address the challenges posed by technical noise and dropout artifacts in transcriptomic data.
• To provide a unified approach for dissecting multi-dimensional tumor heterogeneity.

Main Methods:

• ScSpTITH utilizes a rank-based strategy, selecting highly variable genes and computing pairwise Spearman rank correlations.
• This method is inherently robust to technical noise, non-normality, and high dropout rates.
• The framework is scalable and flexible, applicable to diverse cancer and developmental datasets.

Main Results:

• Elevated ScSpTITH scores correlate strongly with active tumor evolution, advanced stage, and cellular plasticity.
• ScSpTITH effectively quantifies heterogeneity across inter-tumoral and intra-tumoral dimensions, including spatial regions and cell populations.
• The framework demonstrates robustness against high dropout rates common in scRNA-seq data.

Conclusions:

• ScSpTITH provides a stable and interpretable method for quantifying transcriptional heterogeneity.
• The framework links ITH quantification to critical clinical factors such as therapeutic resistance and poor patient outcomes.
• ScSpTITH offers a unified computational solution for analyzing multi-dimensional heterogeneity in single-cell and spatial transcriptomic studies.