Germline Depth-Only

The DRAGEN CNV pipeline also supports a legacy depth-only mode that analyzes read depth coverage to identify copy number variations without using B-allele frequency information. This workflow follows the same processing steps as the ASCN workflow (binning, bias correction, normalization, segmentation, and calling) but operates solely on coverage depth signals.

Key Differences from ASCN Workflow

The depth-only workflow differs from the germline ASCN workflow in the following ways:

• No B-allele analysis: Does not require --cnv-population-b-allele-vcf or generate B-allele count files

• No allele-specific information: Cannot distinguish between copy-neutral LOH and diploid regions, or determine minor allele copy numbers

• Limited output: Does not produce *.ballele.counts.gz, *.baf.bedgraph.gz, *.baf.seg, or cytogenetics VCF files

• VCF format: Uses VCF v4.2 instead of v4.4 by default

• No mosaic detection: Mosaic alteration calling is not available

• Different quality metrics: QUAL scores and filtering thresholds differ from ASCN workflows

QUAL

Quality scores are computed using a probabilistic model that uses a mixture of heavy tailed probability distributions (one per integer copy number) with a weighting for event length. Noise variance is estimated. The output VCF contains a Phred-scaled metric that measures confidence in called amplification (CN > 2 for diploid locus), deletion (CN < 2 for diploid locus), or copy neutral (CN=2 for diploid locus) events.

The scoring algorithm also calculates exact copy-number quality scores that are inputs to the DeNovo CNV detection pipeline.

Example Command Lines

WGS with self-normalization:

dragen \
  -r <HASHTABLE> \
  --output-directory <OUTPUT> \
  --output-file-prefix <SAMPLE> \
  --enable-map-align false \
  --enable-cnv true \
  --bam-input <BAM> \
  --cnv-enable-self-normalization true \
  --sample-sex <SEX>

WES with panel of normals:

dragen \
  -r <HASHTABLE> \
  --output-directory <OUTPUT> \
  --output-file-prefix <SAMPLE> \
  --enable-map-align false \
  --bam-input <BAM> \
  --sample-sex <SEX> \
  --enable-cnv true \
  --cnv-target-bed <CNV_TARGET_BED> \
  --cnv-combined-counts <CNV_PANEL_OF_NORMALS>

Note: To explicitly enable depth-only mode when using DRAGEN v4.5 or later, omit the --cnv-population-b-allele-vcf option.

Depth-Only Specific VCF Annotations

The depth-only (non-ASCN) workflow produces VCF files in VCF v4.2 format with the following specific annotations.

Header

The VCF header is annotated with ##source=DRAGEN_CNV to indicate the file is generated by the legacy depth-only CNV caller.

Available FILTERs

The depth-only workflow applies the following filters to identify potentially false positive CNV calls:

• cnvBinSupportRatio - For CNVs greater than 80kb, indicates the percent span of supporting target intervals is lower than a threshold.

• cnvCopyRatio - Indicates that the segment mean of the CNV is not far enough from copy neutral.

• cnvLength - Indicates that the length of the CNV is lower than a threshold.

• cnvLikelihoodRatio - Indicates a log10 likelihood ratio of ALT to REF is less than a threshold.

• cnvQual - Indicates that the QUAL of the CNV is lower than a threshold.

• dinucQual - Applied based on the percentage of bases in a segment that belong to a two-base set (GC, CT, or AC), determined by individual occurrences. A CNV call is filtered out if any of these percentages fall outside typical ranges, indicating a likely false positive.

• highCN - Indicates a CNV call with implausible copy number (>6). Note: this filter is only applied in WGS depth-only workflows.

INFO Fields

Depth-only workflows include the following INFO fields:

ID	Description
GCP	Percentage of bases that are G or C
CTP	Percentage of bases that are C or T
ACP	Percentage of bases that are A or C

These dinucleotide composition metrics are used by the dinucQual filter to identify false positive calls based on unusual base composition patterns.

FORMAT Fields

The depth-only workflow provides the following common FORMAT fields:

ID	Description
GT	Genotype
SM	Linear copy ratio of the segment mean
CN	Estimated copy number
BC	Number of bins in the region
PE	Number of improperly paired end reads at start and stop breakpoints

Germline depth-only CNV also includes:

ID	Description
LR	Log10 likelihood ratio of ALT to REF

The likelihood ratio is computed using the same probabilistic model described in QUAL. To calculate the likelihood ratio, bins within each segment are evaluated on a per-bin basis. For each bin, the probabilities of amplification (CN > ploidy), deletion (CN < ploidy), and copy-neutral (CN = ploidy) states are calculated. The log10 likelihood ratios are then computed as log10(p(amplification) / p(neutral)) for amplification, and log10(p(deletion) / p(neutral)) for deletion. The segment-level likelihood ratio is obtained by summing the per-bin log10 likelihood ratios across all bins in the segment. For copy-neutral calls, the likelihood ratio is set to 0.

Note that depth-only workflows do not provide allele-specific annotations such as MCN, MCNQ, MCNF, MAF, AS, or OBF, which are exclusive to ASCN workflows.

Segmentation Output Files

The depth-only workflow generates additional segmentation output files:

• *.seg.called — Identical to *.seg with an additional column indicating the initial call (+ for duplication or - for deletion)

• *.seg.called.merged — Similar to *.seg.called with potentially merged segments, plus additional columns for QUAL, FILTER, copy number assignment, ploidy, and improper pairs count

VCF Format Differences

Different workflows or modalities can use a different version of the VCF specs:

• CNV caller: VCF v4.4

• Legacy (depth-only) CNV caller: VCF v4.2

General

Field	VCF v4.2	VCF v4.4
INFO/SVLEN	Positive or Negative	Always Positive

Absence/Loss of Heterozygosity (AOH/LOH) - only present when emitting ASCN calls with legacy VCF v4.2

Field	VCF v4.2	VCF v4.4
ALT	<DEL>,<DUP>	<LOH>
FORMAT/GT	1/2	1/1