EnsembleKQC: An Unsupervised Ensemble Learning Method for Quality Control of Single Cell RNA-seq Sequencing Data

Ma, Anqi; Zhu, Zuolang; Ye, Meiqin; Wang, Fei

doi:10.1007/978-3-030-26969-2_47

Cited by 4 publications

(3 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To reduce the bias caused by the use of arbitrary mtDNA% thresholds, Ma and collaborators (Ma et al 2019) proposed an unsupervised method to optimize the threshold for each given input data. This computationally expensive data-driven procedure, which defines the threshold as a function of the distribution of the data, due to the lack of reference values, is not able to identify bias induced during the library preparation.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Systematic determination of the mitochondrial proportion in human and mice tissues for single-cell RNA sequencing data quality control

Osorio

Cai

2020

Preprint

View full text Add to dashboard Cite

Motivation: Quality control (QC) is a critical step in single-cell RNA-seq (scRNA-seq) data analysis. Data from low-quality cells is removed during the QC process in order to yield genuine, biologically meaningful results. One of the important QC metrics is the mitochondrial proportion (mtDNA%), i.e., the fraction of mitochondrial transcript counts of the total transcript counts. The mtDNA% threshold is used to filter out apoptotic, stressed, low-quality cells in the data. Early publications in the field established a threshold of 5% and since then, it has been used as a default parameter in several software packages for scRNA-seq data analysis. Furthermore, such a single threshold of mtDNA% has been adapted as a standard and used in many scRNA-seq studies. However, the validity of using such a uniform threshold across different species, single-cell technologies, tissues, and cell types has not been adequately assessed. Results: We systematically analyzed 5,530,106 cells reported in 1,349 annotated datasets available in the PanglaoDB database and found that the average mtDNA% in scRNA-seq data across human tissues is significantly higher than that across mouse tissues. This difference is not confounded by single-cell platforms and technologies used to generate the data. Based on this finding, we propose new reference values of the mtDNA% threshold for 121 tissues of mouse and 44 tissues of humans. In general, for mouse tissues, with very few exceptions (3 of 121 analyzed tissues), the 5% threshold can still be used as a default, which performs well to distinguish between healthy and low-quality cells. For human tissues, the 5% threshold should be reconsidered as it fails to accurately discriminate between healthy and low-quality cells in 29.5% (13 of 44) tissues analyzed.

show abstract

Section: Introductionmentioning

confidence: 99%

“…Therefore, having a uniform and standardized threshold for scRNA-seq data analysis is essential. It improves the reproducibility of experiments and simplifies the automatization of bioinformatic pipelines (McCarthy et al 2017).…”

Section: Introductionmentioning

confidence: 99%

Systematic determination of the mitochondrial proportion in human and mice tissues for single-cell RNA sequencing data quality control

Osorio

Cai

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…To avoid the use of arbitrary %mtRNA thresholds, Ma et al suggested an unsupervised method for optimization of quality control parameters, called EnsembleKQC [ 21 ]. The threshold is based on a function of the distribution of the data and represents a more objective method for the quality control of biological samples.…”

Section: Discussionmentioning

confidence: 99%

Quality control in scRNA-Seq can discriminate pacemaker cells: the mtRNA bias

Galow

Kussauer

Wolfien

et al. 2021

Cell. Mol. Life Sci.

View full text Add to dashboard Cite

Single-cell RNA-sequencing (scRNA-seq) provides high-resolution insights into complex tissues. Cardiac tissue, however, poses a major challenge due to the delicate isolation process and the large size of mature cardiomyocytes. Regardless of the experimental technique, captured cells are often impaired and some capture sites may contain multiple or no cells at all. All this refers to “low quality” potentially leading to data misinterpretation. Common standard quality control parameters involve the number of detected genes, transcripts per cell, and the fraction of transcripts from mitochondrial genes. While cutoffs for transcripts and genes per cell are usually user-defined for each experiment or individually calculated, a fixed threshold of 5% mitochondrial transcripts is standard and often set as default in scRNA-seq software. However, this parameter is highly dependent on the tissue type. In the heart, mitochondrial transcripts comprise almost 30% of total mRNA due to high energy demands. Here, we demonstrate that a 5%-threshold not only causes an unacceptable exclusion of cardiomyocytes but also introduces a bias that particularly discriminates pacemaker cells. This effect is apparent for our in vitro generated induced-sinoatrial-bodies (iSABs; highly enriched physiologically functional pacemaker cells), and also evident in a public data set of cells isolated from embryonal murine sinoatrial node tissue (Goodyer William et al. in Circ Res 125:379–397, 2019). Taken together, we recommend omitting this filtering parameter for scRNA-seq in cardiovascular applications whenever possible.

show abstract