Known Limitations with Commercial Controls

Using cell lines or contrived samples for validation or as positive controls may result in inconsistent or unusual results depending on how the samples were constructed. Illumina recommends using commercial controls only for the variant class they are intended for. For example, if the sample is contrived for CNVs but not explicitly for SNV, MSI, TMB, or fusions, we would not recommend using this sample for validation or as a positive control for SNV, MSI, TMB, or fusions.

The table below summarizes known limitations with commercial controls for TruSight Oncology 500 assays.

TruSight Oncology 500 ctDNA v2

Product Name	Variant Class	Description	SW Version Impacted
Seraseq ctDNA Complete Mutation Mix AF0.5% (MN 0710-0531)	Small Variants	Low recall on Deletion chr17 41245586 CT>C (BRCA1 p.K654fs*47) due to a high noise region adjacent to a homopolymer. Under review to be addressed in the future SW versions.	All
Seraseq ctDNA Complete Mutation Mix AF0.5% (MN 0710-0531) or above	All	Lower recall due to the variant observed VAF below the limit of detection (LOD) of the TSO 500 ctDNA assay. See section for more details.	All
Seraseq ctDNA Complete Mutation Mix AF0.5% (MN 0710-0531)	TMB	Not recommended for TMB validation due to many small variants having VAF close to the TMB limit of detection. Small changes in preparation and dilution result in significant changes in TMB scores (behavior not observed in clinical samples).	All
Twist cfDNA Pan-Cancer Reference Standard v2	Fusions	False negatives due to a known issue on high efficiently libraries.	2.6.0 and below
Seraseq ctDNA Mutation Mix v2 (multiple VAF)	Fusions	False negatives due to a known issue on high efficiently libraries.	2.6.0 and below

TruSight Oncology 500 v1 and v2

Product Name	Variant Class	Description	SW Version Impacted
Seraseq® FFPE BRCA1/2 LGR Reference Material (MN 0730-0564)	BRCA LR	Not recommended for LR validation due to 1) high level of noise 2) method of contriving resulting in some artifacts. We have found Seraseq BRCA 1/2 Exon Deletions DNA Mix (MN 0730-0570) to work well.	All
Samples with tumor fraction > 90%	HRD, GIS	The GIS algorithm does not produce reliable results if the sample’s tumor fraction is over 90%, for example, in pure cell samples.	All

Troubleshooting Missing Variants in Control Samples

If you are using commercial controls to test analytical performance of a TSO 500 or TSO 500 ctDNA assay and an expected variant is not reported, use the following troubleshooting steps:

1. Confirm the variant of interest is within the callable assay manifest BED region. Contact Illumina's Technical Support for the manifest file.

2. Confirm the sample passes the QC metric for the relevant variant class.

3. Confirm the truth set and output use the same genome annotation, for example, hg19 truth set to hg19 output.

4. Confirm the variant of interest is expected to be present with allele frequency above the assay's limit of detection. Review recommendations for planning analytical studies for TSO 500 ctDNA.

5. Depending on the control, confirm how variant presence was established. Orthogonal methods can differ from the methods used by TSO 500 or TSO 500 ctDNA.

After these preliminary checks, inspect the intermediate output files for the variant of interest. The relevant files are listed in the DNA output sections but some additional variant specific considerations are below.

• Small variants: If an expected call is not reported, first check whether the variant is present in {SAMPLE_ID}_hard-filtered.vcf.gz. If the variant is present but filtered, review the filter flag to identify the reason. If the variant of interest is not present in that VCF, examine deviations from the reference in {SAMPLE_ID}_hard-filtered.gvcf.gz or visualize the locus in the BAM file.

• CNVs: If an expected call is not reported, review all gene-level fold changes in {SAMPLE_ID}.cnv.vcf. This file reports fold changes regardless of CNV call status. The Segment Mean (SM), which is synonymous with fold change, can be compared directly to the thresholds for amplification and deletion calls used to determine whether a gene is reported as AMP or DEL. In some cases, the fold change is very close to the call threshold, and the lack of a call can be explained by stochastic variability rather than lack of signal. Also note that TSO 500 and TSO 500 ctDNA call only whole-gene deletions or amplifications. If a focal CNV event covers only a small portion of the gene, for example, 1 of 10 exons, the event might not be called.