SciClone: How to interpret Coverage vs VAF plots

Question

[B. Arman Aksoy]

I am currently in the process of trying to understand what each clonality inference does under the cover and while doing that I realize that both the SciClone paper and the R-library feature a multi-panel plot to show Coverage vs VAF (Variant Allele Frequency) nicely aligned with a VAF histogram at the top. I also saw that the same plot type was also featured in a recent review article. However, neither paper gives a rationale regarding the information context for this plot and I would like to better understand what is unique about the VAF/Coverage graph and how it adds to our understanding of clones in addition to what we learn from the VAF distribution.

I am potentially over-interpreting things here, but just wanted to make sure that I don't miss any of the information resources.

Answer 1

[Chris Miller]

The coverage information isn't used in the clustering of VAFs at all (except to establish a minimum threshold). The plot has a couple of nice features though:

1) It can help identify sequencing artifacts (which often present with extremely high coverage and outlier VAFs)

2) it gives some "jitter" to the plots, so that the individual points can be seen in a 2d plot (otherwise they'd all be on a line)

3) Since there will be more sampling error with lower coverages, it helps ground your interpretation of the confidence of the clustering. (higher coverage experiment = more easily separable clusters and higher confidence in the results)

[B. Arman Aksoy]

thanks, Chris — much appreciated!