Comprehensive Noise Reduction in Single-Cell Data with the RECODE Platform

Abstract

AbstractSingle-cell sequencing generates vast amounts of genomic and epigenomic data from thousands of individual cells and can reveal insights into biological principles at the single-cell resolution. However, challenges such as technical noise (dropout) and batch effects hinder obtaining high-resolution structures that are essential for tasks such as the identification of rare cell types and dataset comparison across different cultures. Here, I introduceintegrative RECODE (iRECODE), a comprehensive method for noise reduction that is based on the RECODE platform, which targets the technical noise in single-cell RNA-sequencing data using high-dimensional statistics. I show iRECODE effectively mitigates both technical and batch noise with high accuracy and low computational cost. Additionally, the application of RECODE extended to other single-cell sequencing data types including single-cell Hi-C and spatial transcriptomics data and the recent enhancements in RECODE have markedly improved its accuracy and computational efficiency. Thus, the RECODE platform presents a robust solution for mitigating noise in single-cell sequencing, offering promise for advancing our understanding of biological phenomena beyond transcriptomics, encompassing epigenomic and spatial transcriptomic domains.