CONICS integrates scRNA-seq with DNA sequencing to map gene expression to tumor sub-clones

Abstract

Motivation: Single-cell RNA-sequencing (scRNA-seq) has enabled studies of tissue composition at unprecedented resolution. However, the application of scRNA-seq to clinical cancer samples has been limited, partly due to a lack of scRNA-seq algorithms that integrate genomic mutation data.

Results: To address this, we present.

Conics: COpy-Number analysis In single-Cell RNA-Sequencing. CONICS is a software tool for mapping gene expression from scRNA-seq to tumor clones and phylogenies, with routines enabling: the quantitation of copy-number alterations in scRNA-seq, robust separation of neoplastic cells from tumor-infiltrating stroma, inter-clone differential-expression analysis and intra-clone co-expression analysis.

Availability and implementation: CONICS is written in Python and R, and is available from https://github.com/diazlab/CONICS.

Supplementary information: Supplementary data are available at Bioinformatics online.

Fig. 1.

An example of CONICS analysis on scRNA-seq and exome-seq of a glioblstoma biopsy. (A) CONICS quantifies CNVs in single cells with a controlled error rate: ScRNA-seq read-count correlations with CNV status (top left); scRNAseq read-count distributions for an example CNV (top right); FDR estimates in assigning CNV status to individual cells, computed via cross validation (bottom left) and via comparison to a control dataset (bottom right). (B–D) CONICS estimates CNV allele frequency, which, when compared to the expression of canonical markers and clustering of CNV status, enables the rigorous separation of stromal /immune cells from neoplastic cells. (E) Co-expression network of PTEN, produced by CONICS, compared between cells with a chr. 10 loss and wild-type