Correction: Corrigendum: Characterizing noise structure in single-cell RNA-seq distinguishes genuine from technical stochastic allelic expression

Abstract

Nature Communications 6: Article number: 8687 (2015); Published: 22 October 2015; Updated: 11 January 2016. The original version of this Article contained an error in the spelling of the author Tomislav Ilicic, which was incorrectly given as Tomislav Illicic. This has now been corrected in both the PDF and HTML versions of the Article.

Design and computational analysis of single-cell RNA-sequencing experiments

“…We have focused primarily on analyses for which multiple tools are available, although we note that there are additional methodological challenges that are not highlighted here. For example, novel methods are beginning to be developed for studies of allele-specific expression [ 63 ] and isoform usage [ 87 ]. Methodological challenges are also introduced with the advent of technologies that increase sample size by allowing for routine profiling of tens of thousands of cells [ 61 , 62 , 88 ].…”

“…Capture efficiency (percentage of mRNA molecules in the cell lysate that are captured and amplified), amplification bias (non-uniform amplification of transcripts), and sequencing efficiency (rate at which cDNAs in a library are sequenced) are major contributors to technical variation. These sources affect counts in both a gene- and a cell-specific manner and are observed to have the greatest effect on lowly expressed genes [ 48 , 63 , 64 ]. Considerable variation also results from differences among cells in cell-cycle stage or cell size, variation that is not typically observed in (unsynchronized) bulk RNA-seq experiments in which expression is profiled on average over thousands of cells.…”

“…One group of methods aims to estimate technical variability, with the goal of identifying genes that have overall variability that greatly exceeds that expected from technical sources [ 48 , 54 , 63 ]. These methods use spike-ins to estimate technical noise because spike-ins are exposed to most of the same experimental steps as endogenous genes but are free of biological variation.…”

“…By modeling this relationship, estimates of technical variability are obtained and genes whose expression variability greatly exceeds these estimates for a given biological variability threshold can be identified. Although useful, this approach does not fully capture cell-to-cell differences in technical variability [ 63 ] or give explicit estimates of biological variability [ 9 ]. More recent methods provide improvements by estimating biological variability [ 9 ] or by incorporating additional aspects of technical noise to estimate parameters that account for variation across cells using spike-ins [ 63 ] or jointly over spike-ins and genes [ 54 ].…”

“…These parameters are then used to estimate technical variability for endogenous genes and to test whether each gene exceeds a variability threshold Spike-ins and endogenous genes are normalized separately using the median normalization method. Gene specific P values are provided to identify highly variable genes Kim et al [ 63 ] Uses spike-ins to estimate parameters related to technical variance, allowing for differences in variability across cells. Estimates gene-specific biological variability by subtracting technical variability from total variance Normalization factors are estimated using the median normalization method.…”