Methods for Copy Number Aberration Detection from Single-cell DNA Sequencing Data

Abstract

Background: Accurate detection of copy number aberrations (CNA) can aid in understanding the genetic causes of diseases. Three methods (CopyNumber, Ginkgo, and HMMcopy) have been applied to single-cell DNA sequencing data for CNA detection. Results:In this paper, we benchmarked these three methods on simulated as well as biological datasets. We found that HMMcopy has the best accuracy of the three methods in terms of breakpoint detection but that Ginkgo is better in terms of detecting the actual copy numbers. In terms of computational requirements, HMMcopy consumes the least memory, but is in between the two other methods in terms of running time.Conclusion: While single-cell DNA sequencing is very promising for elucidating and understanding CNAs, even the best existing method does not exceed 80% accuracy. New methods that significantly improve upon the accuracy of these three methods are needed. Furthermore, with the large datasets being generated, the methods must be computationally efficient.

SCOPE: A Normalization and Copy-Number Estimation Method for Single-Cell DNA Sequencing

“…Instead of applying the default locally estimated scatterplot smoothing (LOESS) regression to reduce GC content bias, it adopts a modal regression algorithm that normalizes bin counts to integer values, as expected of single-cell profiles. However, as we show later through benchmark analysis, the enhanced version of HMMcopy suffers from low stability-a finding that concords with another recent benchmark report (Fan et al, 2019).…”

“…Overall, when compared with Ginkgo and HMMcopy, SCOPE returned the highest precision rates, recall rates, joint F-measures (i.e., geometric means of precision and recall), Matthew's correlation coefficients, and the Kappa statistics under all simulation settings (Figure 4B; Table S3). In several cases, HMMcopy returned copy-number estimates that were inflated across the entire genome (Table S3A), and in concordance with a recent benchmark report (Fan et al, 2019), HMMcopy could not correctly predict the absolute copy numbers in the absence of intermediate copy numbers (Figure 4B; Table S3C).…”

Single-cell copy number calling and event history reconstruction

“…SCICoNE also performs well in comparison to HMMcopy [26] (Figure 4, red) which performs copy number calling per cell with a hidden Markov model, and has been found to have good overall performance in single-cell copy number calling [28]. However, for the simulated data with an average read depth of 2-8 (comparable to 10x Genomics data), HMMcopy performs worse than assuming no CNAs occurred (Figure 4, orange).…”