Precision of loss of heterozygosity detection with Sanger sequencing vs copy number analysis

Question

[user31888]

I first identified some somatic mutants with NGS. Then I validated them with Sanger sequencing. With NGS I also looked for loss of heterozygosity (LOH) in my tumor samples using a copy number analysis (CNV) approach (read count based). Some of the variants show LOH with the CNV analysis but not on the corresponding Sanger chromatogram.

From a quantitative point of view, which method (CNV or Sanger) would be the most accurate to detect LOH? Are the peaks on a Sanger chromatogram precise enough to asses the proportion of ref. vs alt. allele?

Thanks !

Answer 1

[Eric T.]

To assess LOH in a tumor sample I think the "gold standard" is a SNP array, but it should be possible to get the equivalent from NGS.

With the data you have, I'd call SNVs from the NGS data and look at the allele frequencies in the regions of interest. If the tumor sample is bulk tissue, the allele frequencies in a region with LOH would be shifted off from .5, but assuming the sample is not fully clonal (hard to manage even with cell lines), they wouldn't be 1.0 either. I'm not sure whether this shift is easy to see in a Sanger chromatogram.

Many copy number callers can use the SNV allele frequencies alongside absolute copy number to infer LOH. In CNVkit, for example, once you have the SNV calls in VCF format you can infer LOH with the 'call' command, and/or plot them with the 'scatter' command to visually evaluate the evidence for LOH.