Author Correction: The shaky foundations of simulating single-cell RNA sequencing data

Abstract

Full text

Correction: Genome Biol 24, 62 (2023)

https://doi.org/10.1186/s13059-023-02904-1

Following publication of the original article [1], it was pointed out that the legend to Table 1 did not match the content of the table.

The incorrect Table 1 is as follows: Table 1. Overview of scRNA-seq simulators compared in this study. Methods are ordered alphabetically and annotated according to their (in)ability to accommodate multiple batches and/or clusters, support for parallelization (parameter estimation and data simulation, respectively), software availability, and publication year. The right-most column catalogues neutral benchmark studies where each simulator was used. (✓ = yes, ✘ = no, (✓) = yes, but based on user input parameters, i.e., no support for parameter estimation, requires random splitting of cells into two groups, †/‡ = internal/prior resampling from empirical parameter distribution, ∘ = no separate estimation step)**BatchesClusters**Type(s)Cell #ParallelizationAvailabilityYearModelBASiCS [37]✓✘b✘✓✘R/Bioc2015NBESCO [38]✓✓n,b,k✓✓✓R/GitHub2020Gamma-Poissonhierarchicell [39]✓✘n,b✓✘✘R/GitHub2021NBmuscat [40]✓✓n,b,k(✓)†✘✘R/Bioc2020NBPOWSC [41]✘✓n,k(✓)†✘✘R/Bioc2020zero-inflated, log-normal Poisson mixturepowsimR [42]✘(✓)n(✓)†✓✓R/GitHub2017NBscDD [43]✘✘n*✓✓✓R/Bioc2016Bayesian NB mixture modelscDesign [44]✘(✓)n✓∘✓R/GitHub2019Gamma-Normal mixture modelscDesign2 [45]✘✓n,k✓✓✘R/GitHub2020(zero-inflated) Poisson or NB + Gaussian copula for gene–gene correlationsSCRIP [46]✓✓n,b,k✓✘✘R/GitHub2020(Beta-)Gamma-PoissonSPARSim [47]✓✘n,b(✓)‡✘✘R/GitLab2020Gamma-multivariate hypergeometricsplatter [15] (Splat model)(✓)(✓)n✓✘✘R/Bioc2017Gamma-PoissonSPsimSeq [16]✓✘n,b✓∘✘R/Bioc2020log-linear model-based density estimation + Gaussian copula for gene–gene correlationsSymSim [48]✓✘n,b✓✘✘R/GitHub2019kinetic model using MCMCZINB-WaVE [49]✓✓n,b,k✘✘✘R/Bioc2018zero-inflated NBzingeR [50]✘✘n(✓)†‡✘✘R/GitHub2017zero-inflated NB

The corrected Table 1 is as follows:

Table 1. Overview of scRNA-seq simulators compared in this study. Methods are ordered alphabetically and annotated according to their (in)ability to accommodate multiple batches and/or clusters, support for parallelization (parameter estimation and data simulation, respectively), software availability, and publication year. ‘Type(s)’ column specifies which type of simulations can be produced (n: “singular” references: single batch or cluster; b: multiple batches; k: multiple clusters). ‘Cell #’ refers to whether the number of cells can be varied. Symbols: ✓ = yes, ✘ = no, (✓) = yes, but based on user input parameters, i.e., no support for parameter estimation, requires random splitting of cells into two groups, †/‡ = internal/prior resampling from empirical parameter distribution, ∘ = no separate estimation step)**BatchesClusters**Type(s)Cell #ParallelizationAvailabilityYearModelBASiCS [37]✓✘b✘✓✘R/Bioc2015NBESCO [38]✓✓n,b,k✓✓✓R/GitHub2020Gamma-Poissonhierarchicell [39]✓✘n,b✓✘✘R/GitHub2021NBmuscat [40]✓✓n,b,k(✓)†✘✘R/Bioc2020NBPOWSC [41]✘✓n,k(✓)†✘✘R/Bioc2020zero-inflated, log-normal Poisson mixturepowsimR [42]✘(✓)n(✓)†✓✓R/GitHub2017NBscDD [43]✘✘n*✓✓✓R/Bioc2016Bayesian NB mixture modelscDesign [44]✘(✓)n✓∘✓R/GitHub2019Gamma-Normal mixture modelscDesign2 [45]✘✓n,k✓✓✘R/GitHub2020(zero-inflated) Poisson or NB + Gaussian copula for gene–gene correlationsSCRIP [46]✓✓n,b,k✓✘✘R/GitHub2020(Beta-)Gamma-PoissonSPARSim [47]✓✘n,b(✓)‡✘✘R/GitLab2020Gamma-multivariate hypergeometricsplatter [15] (Splat model)(✓)(✓)n✓✘✘R/Bioc2017Gamma-PoissonSPsimSeq [16]✓✘n,b✓∘✘R/Bioc2020log-linear model-based density estimation + Gaussian copula for gene–gene correlationsSymSim [48]✓✘n,b✓✘✘R/GitHub2019kinetic model using MCMCZINB-WaVE [49]✓✓n,b,k✘✘✘R/Bioc2018zero-inflated NBzingeR [50]✘✘n(✓)†‡✘✘R/GitHub2017zero-inflated NB

The linked citations and the hyperlinks to the availability of data (in table 1) can be found in the original article. The original article [1] has been corrected.