ATAXIC: An Algorithm to Quantify Transcriptomic Perturbation Heterogeneity in Single Cancer Cells

Abstract

The single-cell RNA sequencing (scRNA-seq) has recently been widely utilized to quantify transcriptomic profiles in single cells of bulk tumors. The transcriptomic profiles in single cells facilitate the investigation of intratumor heterogeneity that is unlikely confounded by the nontumor components. We proposed an algorithm (ATAXIC) to quantify the heterogeneity of transcriptomic perturbations (TPs) in single cancer cells. ATAXIC calculated the TP heterogeneity level of a single cell based on the standard deviations of the absolute z-scored gene expression values for tens of thousands of genes, reflecting the asynchronous degree of transcriptomic alterations relative to the central (mean) tendency. By analyzing scRNA-seq datasets for eight cancer types, we revealed that ATAXIC scores were likely to correlate positively with the enrichment scores of various proliferation and oncogenic signatures, DNA damage repair, treatment resistance, and unfavorable phenotypes and outcomes in cancer. The ATAXIC scores varied among different cancer types, with lung cancer and melanoma having the lowest average scores and clear cell renal cell carcinoma having the highest average scores. The low TP heterogeneity in lung cancer and melanoma could bestow relatively higher response rates to immune checkpoint inhibitors on both cancer types. In conclusion, ATAXIC is a useful algorithm to quantify the TP heterogeneity in single cancer cells, as well as providing new insights into tumor biology.