Mosaic Genome

A mosaic genome describes an individual whose cells harbor two or more distinct genetic sequences, arising from postzygotic mutations during development. This guide is for researchers, clinicians, and bioinformatics trainees who need a practical framework to identify, interpret, and report mosaic variants. It draws on authoritative sources such as the NCBI Bookshelf NCBI Bookshelf and EMBL-EBI Training EMBL-EBI Training to ground every step in best practices.

Mosaicism is not a rare phenomenon. It underlies diverse conditions from neurodevelopmental disorders to cancer and even normal aging. Understanding its detection limits and interpretation pitfalls is essential. The Galaxy Training Network Galaxy Training Network offers hands on workflows that demonstrate why careful read depth and allele frequency thresholds matter.

At a Glance

Core Concepts

A mosaic genome arises when a mutation occurs after fertilization, producing a patchwork of cells with different genotypes. The mutation may be present in a single cell (very low frequency) or in a large clone (detectable by standard sequencing). Two main categories exist: somatic mosaicism, confined to non-reproductive tissues, and germline mosaicism, where the mutation appears in a subset of gametes. The NCBI Bookshelf NCBI Bookshelf provides foundational descriptions of these mechanisms, including the distinction between de novo and mosaic variants.

The timing of the mutation determines its distribution. An early embryonic mutation can affect many tissues and may appear in a proportion of cells across the body. Later mutations remain restricted to a smaller lineage. The landscape of parental postzygotic mutations was recently characterized in a study of over 11,000 rare disease trios Landscape of parental postzygotic mutations across >11,000 rare disease trios, revealing that approximately 5% of apparently de novo variants are actually mosaic in a parent. This finding underscores the importance of considering mosaicism in clinical genetic testing.

Detection relies on sequencing depth. Standard germline variant calling (30,50x coverage) rarely picks up alleles present in fewer than 20% of cells. Dedicated mosaic variant calling pipelines require depth of 100x or more, plus specialized statistical models to distinguish true low-frequency variants from sequencing error. The EMBL-EBI Training EMBL-EBI Training modules on variant analysis cover these depth requirements and the trade-offs between sensitivity and false positive rate.

Decision Points

When should you suspect a mosaic genome? The following decision criteria guide when to pursue dedicated mosaic analysis:

• Variant allele frequency (VAF) much lower than 50%. A variant present in 10,30% of reads in a diploid sample (and not artefactual) suggests mosaicism.
• Recurrence of a variant in multiple tissues but at different frequencies. For example, a mutation found in skin fibroblasts but absent in blood.
• A variant that appears de novo in an offspring but is undetectable in parental blood. This pattern can indicate parental germline mosaicism.
• Unexpected phenotypic severity or discordance between genotype and phenotype. Mosaicism for a pathogenic variant can explain milder or atypical presentations.

The Bioconductor project Bioconductor provides R packages (e.g., MosaicCBR, MosaicFinder) that implement statistical tests to differentiate true mosaic calls from background noise. A practical decision tree appears in the Galaxy Training Network tutorials: if the depth is lower than 80x, do not attempt to call variants below 5% VAF, if depth is 100,200x, use a Bayesian model that incorporates base quality and strand bias.

Another critical decision point involves assessing the clinical context. In rare disease trios, the study referenced above Landscape of parental postzygotic mutations across >11,000 rare disease trios showed that parental mosaicism explains about 5% of cases where a child appears to carry a de novo mutation. When a recurrence risk for a family is needed, testing multiple tissues from both parents increases the chance of detecting low-level mosaicism.

Practical Workflow

The following workflow outlines steps to identify and evaluate mosaic variants using publicly available data and tools. Adapt these steps to your sequencing platform and research question.

1. Data access. Retrieve raw sequencing reads from the NCBI Sequence Read Archive NCBI Sequence Read Archive or from your own sequencing run. Ensure that the data is paired end and that the read length is at least 100 base pairs for reliable mapping.

2. Quality control and preprocessing. Run FastQC and MultiQC to check for adapter contamination and base quality. Trim low quality bases with Trimmomatic or cutadapt. Output fastq files should pass standard QC metrics.

3. Read alignment. Use BWA MEM to align reads to the reference genome. For human samples use GRCh38. Mark duplicates with Picard. Quality metrics such as mean depth and coverage uniformity should be recorded.

4. Mosaic variant calling. Use specialized tools such as GATK Mutect2 (in tumor only mode with the ,genotype-germline-sites flag) or VarScan2 with a minimum VAF of 0.01. The Galaxy Training Network Galaxy Training Network provides a complete workflow that includes recalibration of base qualities and downstream filtering. Run with high stringency for base quality (minimum Q30) and read mapping quality (minimum Q20).

5. Filter candidates. Remove variants with strand bias (Fisher’s exact test p > 0.05), low read depth (< 20 reads at site), or high background noise in nearby regions. Keep only variants that appear in at least 5 independent reads.

6. Validation by orthogonal method. For high confidence calls, design a targeted deep amplicon sequencing assay (e.g., 10,000x depth) or use digital PCR. The case report of generalized nevus lipomatosis Generalized Nevus Lipomatosus Cutaneous Superficialis: A Case Report with Comprehensive Genetic Analysis used Sanger sequencing and droplet digital PCR to confirm a mosaic PTEN variant in affected skin tissue.

7. Interpretation. Annotate variants using Ensembl VEP or SnpEff. Check population frequencies in gnomAD (mosaic variants rarely appear at appreciable frequencies). Assess pathogenicity using ClinVar, and consider the tissue distribution of the variant.

Quality Checks

Quality checks for mosaic genome analysis differ from germline analysis because the signal to noise ratio is much lower. The following metrics should be monitored:

• Base quality scores. Mean base quality at variant position should be >= 30.
• Mapping quality. Reads supporting the alternate allele must have mapping quality >= 20.
• Strand bias. The alternate allele should be present on both forward and reverse strands. A Fisher’s exact test for strand bias should yield a p value above 0.05.
• Read depth. Minimum 100x total depth at the site for calls below 5% VAF. The NCBI Bookshelf chapter on sequencing depth recommendations NCBI Bookshelf notes that 200x depth can detect variants at 2% VAF with 80% sensitivity.
• Allele frequency distribution in surrounding region. A true mosaic variant will not cluster with many other low frequency variants in a small window, which would suggest a mapping artefact.
• Cross sample contamination. Estimate contamination using VerifyBamID. Contamination above 3% can create false positive low frequency calls.

Use the Bioconductor Bioconductor package MosaicQC to automate these checks. The Galaxy Training Network workflow includes a quality report that flags samples with excessive heterozygosity or abnormal VAF distributions.

Common Mistakes

Overinterpreting low-frequency variants remains the most frequent error. A VAF of 1% can easily arise from sequencing errors in regions with repetitive sequences, even after stringent filtering. Always require confirmation with an independent method. The piggyBac transgenesis platform study Highly efficient and low-mosaicism piggyBac transgenesis platform for rapid founder phenotyping emphasizes that mosaic rates below 5% require extremely high read depths to differentiate from background.

Ignoring tissue specific mosaicism leads to misclassification. A variant present in blood but absent in skin might be dismissed as an artefact, yet it could represent a genuine hematopoietic mosaic. Conversely, failure to sample the affected tissue reduces sensitivity. The case of generalized nevus lipomatosis Generalized Nevus Lipomatosus Cutaneous Superficialis: A Case Report with Comprehensive Genetic Analysis illustrates that the pathognomonic variant was only detectable in a skin biopsy, not in blood.

Using default germline variant callers for mosaic detection is another common mistake. Tools like GATK HaplotypeCaller assume a VAF near 50% or 100% and will often filter out genuine low frequency variants. Always use tools designed for somatic or mosaic calling.

Limits of Interpretation

Mosaic genome analysis has inherent limits that must be acknowledged.

• Detection threshold. Even with deep sequencing, a mutation present in less than 1% of cells is usually undetectable. For a 200x depth run, the minimum detectable VAF is about 2% given standard error models. Lower frequencies require single cell sequencing or extreme depth (10,000x) which is not practical for most studies.
• Tissue sampling bias. A negative result from blood does not rule out mosaicism in other tissues. The “landscape” study Landscape of parental postzygotic mutations across >11,000 rare disease trios found that parental mosaicism was detectable in blood only 57% of the time when it was present in gametes.
• Inability to distinguish from somatic mutations in non model organisms. For example, in the evolutionary study of multidrug resistant Salmonella Evolutionary dynamics of the multidrug-resistant Salmonella Infantis harbouring the pESI megaplasmid across Europe, plasmid mosaicism (variable copy numbers) confounded standard variant calling. In plants such as durum wheat Durum Wheat cv. Svevo Reference Genome Rel.2.0: A Comprehensive Tool for Wheat Genomics, genome wide mosaicism from polyploidy complicates interpretation.
• Germline versus postzygotic discrimination. Without parental samples, a mosaic variant cannot be reliably distinguished from a low frequency germline heterozygous variant. The presence of the variant in multiple tissues is the strongest evidence for postzygotic origin.
• Recombinant forms. In viruses like HIV, circulating recombinant forms (CRFs) represent mosaic genomes at the population level Sequence Notes: Identification of a Novel HIV-1 Circulating Recombinant Form (CRF200_0755) in a Southern Border Region of Yunnan, China. These are not somatic mosaicism but genomic chimerism from coinfection, and the analytical approach differs accordingly.

Frequently Asked Questions

What is the difference between a mosaic variant and a de novo variant?
A de novo variant is present in every cell of an individual and absent in both parents. A mosaic variant is present in only a subset of cells. Distinguishing them requires high depth sequencing of the proband and often multiple tissues or parental samples.

How can I detect germline mosaicism in a parent?
Sequence the parent’s blood and, if possible, other tissues such as saliva or skin. If a child’s de novo variant appears at very low VAF (e.g., 1,5%) in the parent’s blood, it likely represents low level mosaicism. The study of >11,000 trios used deep sequencing of parental blood and detected parental mosaicism in 5% of cases.

Is whole genome sequencing better than exome sequencing for mosaic detection?
Whole genome sequencing covers intronic and intergenic regions where mosaic mutations can occur, and it often has more uniform depth across the genome. However, exome sequencing at 200x depth can still detect mosaic coding variants. Both methods require high depth and dedicated pipelines.

Can mosaic variants cause cancer?
Yes. Somatic mosaicism for driver mutations can lead to clonal expansion and tumor formation. For example, mosaic TP53 mutations in normal skin are associated with increased cancer risk. The analytical framework described here also applies to tumor sequencing, although tumor samples often have higher VAF from clonal expansion.

References and Further Reading

• NCBI Bookshelf: Free textbooks on genetics and sequencing technology. NCBI Bookshelf
• EMBL-EBI Training: Courses on variant calling and molecular biology. EMBL-EBI Training
• Galaxy Training Network: Hands on workflow for mosaic variant detection. Galaxy Training Network
• Bioconductor: R packages for genomic analysis including mosaic detection. Bioconductor
• NCBI Sequence Read Archive: Repository for raw sequencing data. NCBI Sequence Read Archive
• Landscape of parental postzygotic mutations across >11,000 rare disease trios. Am J Hum Genet (2025). PubMed
• Evolutionary dynamics of multidrug-resistant Salmonella Infantis harbouring pESI megaplasmid. Microb Genom (2025). PubMed
• Durum Wheat cv. Svevo Reference Genome Rel.2.0. Plant Biotechnol J (2025). PubMed
• Generalized Nevus Lipomatosus Cutaneous Superficialis: Genetic analysis. Appl Clin Genet (2025). PubMed
• Highly efficient and low-mosaicism piggyBac transgenesis platform. iScience (2025). PubMed
• Identification of novel HIV-1 circulating recombinant form CRF200_0755. AIDS Res Hum Retroviruses (2025). PubMed

Related Articles

• Protein Synthesis
• Incomplete Dominance Gene
• Cell Membrane Function Biology
• Dna Structure
• Protein Structure