Overview

Copy number variants (CNVs) are detected using DRAGEN CNV with self-normalisation and the Shifting Level Models (SLM) segmentation mode. High quality CNVs >10 kb in size are defined as those detected by DRAGEN CNV with filter status 'PASS'. CNVs between 2 and 10 kb in size are identified by combining the results of DRAGEN CNV and DRAGEN SV callers. CNVs in this range detected by both callers with a minimum reciprocal overlap of 50% are deemed to be high quality. The Genomics England Rare Disease Interpretation Pipeline currently annotates and reports all high quality CNVs that overlap with gene biotypes considered during tiering and are ≥ 2 kb in size. Annotations include internal allele frequencies calculated for 5,415 samples.

An overview of the CNV tiering pipeline can be seen below.

Note: there are two separate streams of analysis applied during tiering of CNVs (both use information available in PanelApp: (1) overlap of CNV regions, and (2) overlap of genes; as indicated by different colour boxes in diagram). If a CNV is assigned as Tier A through either stream of analysis, then this is the information that is presented in the interpreted genome output and in the interpretation portal

Only CNV calls from the proband are annotated and displayed. CNV calls in relatives are not currently considered in tiering. However, visual assessment of CNVs for all family members can be performed using coverage profiles displayed in the IGV viewer following the links from the Interpretation Portal.

Tier A

A CNV is assigned Tier A if it satisfies one or both of the following criteria:

• The CNV must not exceed the frequency threshold for both of the CNV frequency metrics (CNV_AF, CNV_AUC). The thresholds are unique to the CNV type:

  • LOSS: 0.005
  • GAIN: 0.01

Note

A CNV can still be included as Tier A if one of the CNV frequency metrics (e.g. CNV_AUC) is exceeded but the other is below the threshold (e.g. CNV_AF)

• The CNV overlaps with a green gene in a panel applied in the analysis. All CNVs (i.e. loss and gain) overlapping any gene are considered, without requiring a minimum overlap threshold.
• The CNV overlaps a pathogenic region in a virtual panel applied in the analysis, the overlap is above the threshold defined in PanelApp for that region, and the variant type matches (i.e., loss or gain) that of the region in PanelApp.

Note

If a Tier A CNV impacts a gene with a biallelic mode of inheritance then the prioritisation algorithm will include assessment of compound heterozygous variants across variant types.

Tier Null

All PASS variants that overlap genes with the biotypes included below, that do not satisfy the Tier A conditions.

Note

Mode of inheritance is not considered in CNV tiering. In contrast to small variant tiering, a single heterozygous CNV within or impacting a gene or region (with appropriate variant type and overlap) will be tiered regardless of the expected mode of inheritance recorded in PanelApp.

CNV biotypes considered during variant tiering

CNVs will be considered during variant tiering if they impact genes with any of the following biotypes:

Biotype	Description
IG_C_gene IG_D_gene IG_J_gene IG_V_gene TR_C_gene TR_D_gene TR_J_gene TR_V_gene	Immunoglobulin (Ig) variable chain and T-cell receptor (TcR) genes imported or annotated according to the IMGT.
protein_coding	Contains an open reading frame (ORF).
nonsense_mediated_decay	If the coding sequence (following the appropriate reference) of a transcript finishes >50bp from a downstream splice site then it is tagged as NMD. If the variant does not cover the full reference coding sequence then it is annotated as NMD if NMD is unavoidable i.e. no matter what the exon structure of the missing portion is the transcript will be subject to NMD.
non_stop_decay	Transcripts that have polyA features (including signal) without a prior stop codon in the CDS, i.e. a non-genomic polyA tail attached directly to the CDS without 3' UTR. These transcripts are subject to degradation.