Accurate and interpretable gene expression imputation on scRNA-seq data using IGSimpute

Abstract

Single-cell RNA-sequencing (scRNA-seq) enables the quantification of gene expression at the transcriptomic level with single-cell resolution, enhancing our understanding of cellular heterogeneity. However, the excessive missing values present in scRNA-seq data (termed dropout events) hinder downstream analysis. While numerous imputation methods have been proposed to recover scRNA-seq data, high imputation performance often comes with low or no interpretability. Here, we present IGSimpute, an accurate and interpretable imputation method for recovering missing values in scRNA-seq data with an interpretable instance-wise gene selection layer. IGSimpute outperforms ten other state-of-the-art imputation methods on nine tissues of the Tabula Muris atlas with the lowest mean squared error as the chosen benchmark metric. We demonstrate that IGSimpute can give unbiased estimates of the missing values compared to other methods, regardless of whether the average gene expression values are small or large. Clustering results of imputed profiles show that IGSimpute offers statistically significant improvement over other imputation methods. By taking the heart-and-aorta and the limb muscle tissues as examples, we show that IGSimpute can also denoise gene expression profiles by removing outlier entries with unexpected high expression values via the instance-wise gene selection layer. We also show that genes selected by the instance-wise gene selection layer could indicate the age of B cells from bladder fat tissue of the Tabula Muris Senis atlas. IGSimpute has linear time-complexity with respect to cell number, and thus applicable to large datasets.

Cofea: correlation-based feature selection for single-cell chromatin accessibility data

“…The development of feature selection methods has been mature for scRNA-seq data. These methods can be divided into three major categories: (i) approaches based on genetic variance across all cells (12); (ii) approaches based on Gini index of gene expression (13,14); and (iii) approaches based on the relationship between dropout events and gene expression (15,16). By contrast, methods dedicated to scCAS data are still in their infancy, and the close-to-binary nature of scCAS data impede the direct applicability of methods for scRNA-seq data to scCAS data.…”