Origins and characterization of variants shared between databases of somatic and germline human mutations

William Meyerson, John Leisman, Fabio C. P. Navarro, Mark GersteinBMC Bioinformatics 2020 · DOI · OpenReview

Abstract

Mutations arise in the human genome in two major settings: the germline and the soma. These settings involve different inheritance patterns, time scales, chromatin structures, and environmental exposures, all of which impact the resulting distribution of substitutions. Nonetheless, many of the same single nucleotide variants (SNVs) are shared between germline and somatic mutation databases, such as between the gnomAD database of 120,000 germline exomes and the TCGA database of 10,000 somatic exomes. This study seeks to explain this overlap. After strict filtering, we found 336,987 variants shared between the somatic and germline databases. A uniform statistical model explains 34% of these shared variants; a model incorporating varying mutation rates of basic mutation types explains another 50%; and a model including extended nucleotide contexts explains an additional 4%. Overall, we find that shared variants largely represent bona fide biological occurrences of the same variant in both settings, arising primarily because DNA has similar basic chemical vulnerabilities in either environment.

Introduction

In genomics, germline mutations (inherited) and somatic mutations (acquired during a lifetime) are typically treated as distinct categories. However, large-scale databases like the Genome Aggregation Database (gnomAD) and The Cancer Genome Atlas (TCGA) share a surprising number of identical variants. This study investigates whether this overlap is a result of statistical chance, technical artifacts like "leakage" (misclassified germline variants in somatic call sets), or shared biological mechanisms.

Methods

The researchers developed a rigorous pipeline to analyze the overlap between 120,000 germline exomes and 10,000 somatic exomes.

  1. Strict Filtering: A "clean" list of variants was created by excluding common polymorphisms, regions with poor sequencing coverage, and sites with low mappability.
  2. Statistical Modeling: The Forbes coefficient of association was used to measure enrichment. A partition-conditioned Forbes model was introduced to determine how much of the overlap could be explained by specific factors, such as basic mutation types (e.g., C→T transitions) or extended nucleotide contexts (k-mers).
  3. Contextual Analysis: The study examined "k-mers" (sequences of 3, 5, or 7 bases) to see if the local genomic neighborhood influenced the likelihood of a mutation being shared.
  4. Leakage Estimation: By analyzing matched normal read depths, the authors attempted to quantify how many somatic calls were actually contaminating germline signals.

Results

The analysis identified 336,987 shared variants after strict filtering. These variants occur nearly three times more often than would be expected by random chance, indicating a significant statistical enrichment.

Statistical association and mutation rate correlations. (A) Distribution of the Forbes coefficient across samples. (B) Correlation between somatic and germline mutation rates stratified by heptamer GC content.

Explanatory Power of Models

The study found that the vast majority of the overlap is predictable:

  • 34% is explained by statistical independence.
  • 50% is explained by basic mutation types (transitions and transversions).
  • 4% is explained by extended nucleotide context (k-mers).
  • 9% remains unexplained, potentially due to shared epigenetic features or unknown technical factors.

Shared Vulnerability and Evolutionary Age

The findings suggest that DNA is chemically identical in both settings and "breaks" in the same ways. Cancers driven by endogenous, "clock-like" mutational processes (Signature 1) show the highest overlap with germline databases. Furthermore, recently derived germline variants and de novo mutations are significantly more likely to be shared with somatic databases than ancient ones, suggesting that natural selection filters germline mutations over millennia in ways that somatic mutations have not yet experienced.

Discussion and Conclusion

The study concludes that shared variants largely represent bona fide biological occurrences rather than sequencing errors. While there is a persistent concern regarding genomic privacy—specifically that somatic databases might "leak" germline data—the authors estimate an upper bound of only 2.3 leaked germline variants per somatic exome.

Ultimately, the similar chemical vulnerability of DNA between somatic and germline settings can be leveraged in future studies to identify disease-related genes. By using germline mutation patterns to inform background-mutation models in cancer, researchers can more accurately distinguish rare driver mutations from the surrounding genomic noise.

Summary updated

Abstract

Background Mutations arise in the human genome in two major settings: the germline and the soma. These settings involve different inheritance patterns, time scales, chromatin structures, and environmental exposures, all of which impact the resulting distribution of substitutions. Nonetheless, many of the same single nucleotide variants (SNVs) are shared between germline and somatic mutation databases, such as between the gnomAD database of 120,000 germline exomes and the TCGA database of 10,000 somatic exomes. Here, we sought to explain this overlap. Results After strict filtering to exclude common germline polymorphisms and sites with poor coverage or mappability, we found 336,987 variants shared between the somatic and germline databases. A uniform statistical model explains 34% of these shared variants; a model that incorporates the varying mutation rates of the basic mutation types explains another 50% of shared variants; and a model that includes extended nucleotide contexts (e.g. surrounding 3 bases on either side) explains an additional 4% of shared variants. Analysis of read depth finds mixed evidence that up to 4% of the shared variants may represent germline variants leaked into somatic call sets. 9% of the shared variants are not explained by any model. Sequencing errors and convergent evolution did not account for these. We surveyed other factors as well: Cancers driven by endogenous mutational processes share a greater fraction of variants with the germline, and recently derived germline variants were more likely to be somatically shared than were ancient germline ones. Conclusions Overall, we find that shared variants largely represent bona fide biological occurrences of the same variant in the germline and somatic setting and arise primarily because DNA has some of the same basic chemical vulnerabilities in either setting. Moreover, we find mixed evidence that somatic call-sets leak appreciable numbers of germline variants, which is relevant to genomic privacy regulations. In future studies, the similar chemical vulnerability of DNA between the somatic and germline settings might be used to help identify disease-related genes by guiding the development of background-mutation models that are informed by both somatic and germline patterns of variation.
BibTeX
@article{meyerson2020origins,
  author = {William Meyerson and John Leisman and Fabio C. P. Navarro and Mark Gerstein},
  title = {{Origins and characterization of variants shared between databases of somatic and germline human mutations}},
  journal = {BMC Bioinformatics},
  year = {2020},
  doi = {10.1186/s12859-020-3508-8},
}
Figure