Molecular and Computational Tools for Identifying Somatic Mosaicism in Human Tissues

Abstract Human genomes harbor significant variation both between and within individuals. Numerous studies have explored inherited variation across human populations and linked various germline polymorphisms to human traits and disease susceptibility. Genomic sequences also vary within an individual, occurring after zygote formation and leading to variation present in a frequency spectrum ranging from individual cells to entire tissues. This somatic mosaicism of genome variation has been well established in cells of phenotypically normal individuals and has been shown to also be associated with some disease phenotypes, particularly cancers. However, these investigations have been mostly limited to higher frequency mosaicism (e.g. >5-10% variant allele frequency) due to technical limitations in both molecular assays and computational methodology. Compounding these technological challenges is that each human tissue exhibits apparently different rates of somatic mosaicism. For example, it is currently estimated that each cell within the human brain contains hundreds to a few thousand somatic single-nucleotide variants (SNVs) and that a smaller fraction of cells harbor somatic copy number variations (CNVs), mobile element insertions (MEIs), and short tandem repeat expansions (STRs). In contrast, somatic mutation rates have been reported to be significantly higher in the large and small intestines and lower in gastric and prostatic glands. These rates have been ascertained through a variety of approaches, including SNP microarrays, bulk and single cell whole genome sequencing, and direct amplification and sequencing of candidate events, each with its own advantages and limitations. However, there has yet to be a systematic investigation of human somatic mosaicism across the entire frequency spectrum within human tissues. Our team has extensive collective experience developing tools for identifying somatic mosaicism in the human brain, including recent surveys of SNV prevalence from whole genome and exome sequencing, CNVs from single cell short-read and nanopore genome sequencing, and retrotransposons through targeted capture. Here, we propose to improve, optimize, and extend our approaches to other human tissues as part of the SMaHT initiative, which will provide an excellent platform for systematically identifying, cataloging, and exploring human somatic mosaicism across tissues. We will achieve this through two phases: in the UG3 phase of this project, we will (1) improve molecular assays for nanopore targeted bulk capture and single cell sequencing and (2) improve computational approaches for detecting somatic mosaicism from single cell and bulk tissue data, while in the UH3 phase we will (3) optimize, benchmark, and validate molecular assays for high- throughput application across human tissues and (4) improve efficiency, runtime, and structured reporting of somatic variants. Collectively, these efforts will enhance our ability to detect at scale previously overlooked classes of somatic variation and extend the size range and frequency spectrum for which they may be ascertained.