Reference cohort selection
Cohort size requirements
Assay Type | Recommended Number of Samples | Minimal Number of Samples* |
WES / Large gene panel | 30-60 | 20 |
Small gene panel (<1 mb)** | 50-100 | 30 |
Somatic gene panel | 50-100 | 30 |
Somatic gene panel + backbone | 50-100 | 30 |
WGS | 30-60 | 15 |
WGS + Homology genes haplotyping | 30-60 | 20 |
WGS - Lowpass | 30-60 | 20 |
* Using fewer than the recommended samples may reduce model quality and increase calling artifacts
** Small gene panels require the sample list file to include each reference sample's sex, particularly when the panel omits sex chromosome genes.
Reference samples selection criteria
Standardize sample preparation - all reference samples must use the same capture kit (for WES and panels), library preparation protocol, and sequencing platform. All future samples run against this model must follow the same protocol.
Biological Diversity:
Samples should come from individuals free of known large chromosomal abnormalities
Avoid enrichment for specific diseases or genetic syndromes
Ideally, samples must be unrelated
Balanced distribution of sample sex (unless the sequencing panel does not include sex chromosomes)
High-quality samples - exclude low- or uneven-coverage samples.
Multiple sample batch - Use samples from multiple preparation batches and sequencing runs to minimize batch bias.
Model configurations
Parameter | Assay Type | Options | Default | Description |
Gene Scope | Somatic | True/False | False | When True, caps the size of called CNVs to a single gene. Use for panels with widely spaced covered regions (non-backbone kits) |
Max Deletion CN | Somatic | 0.1-1.9 | ≤ 1.0 | Maximum copy-number threshold for a valid somatic deletion |
Min Duplication CN | Somatic | 2.1-10 | ≥ 2.8 | Minimum copy-number threshold for a valid somatic duplication |
WGS Homology Haplotyping | WGS | True/False | False | Homology haplotype calling for SMN and HBA |
Mosaic Calling | WGS | True/False | True | Enable mosaic variant calling |
Mosaic Level | WGS | 0.2-0.8 | 0.4 | Mosaic detection sensitivity threshold. At 0.4, only variants with CN ≥ 2.4 or CN ≤ 1.6 are called. |