Comprehensive evaluation of noise reduction methods for single-cell RNA sequencing data

Abstract

Normalization and batch correction are critical steps in processing single-cell RNA sequencing (scRNA-seq) data, which remove technical effects and systematic biases to unmask biological signals of interest. Although a number of computational methods have been developed, there is no guidance for choosing appropriate procedures in different scenarios. In this study, we assessed the performance of 28 scRNA-seq noise reduction procedures in 55 scenarios using simulated and real datasets. The scenarios accounted for multiple biological and technical factors that greatly affect the denoising performance, including relative magnitude of batch effects, the extent of cell population imbalance, the complexity of cell group structures, the proportion and the similarity of nonoverlapping cell populations, dropout rates and variable library sizes. We used multiple quantitative metrics and visualization of low-dimensional cell embeddings to evaluate the performance on batch mixing while preserving the original cell group and gene structures. Based on our results, we specified technical or biological factors affecting the performance of each method and recommended proper methods in different scenarios. In addition, we highlighted one challenging scenario where most methods failed and resulted in overcorrection. Our studies not only provided a comprehensive guideline for selecting suitable noise reduction procedures but also pointed out unsolved issues in the field, especially the urgent need of developing metrics for assessing batch correction on imperceptible cell-type mixing.

The Power of Single-Cell RNA Sequencing in eQTL Discovery

“…Current noise reduction methods for scRNA-seq data include correcting for batch effect and normalization of the sequencing data. A recent study comprehensively analyzed 28 noise-reducing methods and tools in 55 scenarios comprising of real and simulated datasets and proposed a guideline to select suitable procedures [ 38 ]. The study concluded that not a single method can be selected as generalized approach for all scRNA-seq experiments, selection of an appropriate method needs caution and depends on the study design.…”

“…In this case, reciprocal PCA model is recommended. Similarly, linear models are also sensitive to cell population imbalances, and their performance is improved by using cell groups as covariate as in scMerge [ 38 , 40 ]. By unmasking the true biological signals of interest, such methods are expected to also improve the detection of significant cis -eQTL associations in the future.…”

Robust probabilistic modeling for single-cell multimodal mosaic integration and imputation via scVAEIT

“…However, one common limitation of the current approaches is that only simple linear relationships between modalities are captured in the model[2]. And it is hard to incorporate different types of covariates for batch effect adjustment while retaining biological variation [29, 18, 6]. Besides, it is difficult for matrix factorization algorithms to align new datasets without training on the new dataset again.…”

On the importance of data transformation for data integration in single-cell RNA sequencing analysis

“…They developed a set of pipelines for single-cell processing and compared these for various analysis purposes. Their comprehensive benchmark gives the user a practical view of choices [26].…”

“…They developed a set of pipelines for single-cell processing and compared these for various analysis purposes. Their comprehensive benchmark gives the user a practical view of choices [26]. This review of recent benchmark studies highlights the importance of the discussed analysis methods in various tasks.…”