The sensitivity of the Tapestri Platform has been demonstrated to be as low as 0.1 % population frequency. This has been shown with internally conducted cell-line experiments (K562/RAJI spike-in experiments) as well as published data.
Note: 0.1 % population frequency equates to 0.05 % VAF for most cancerous mutations (somatic = heterozygous).
In Tapestri Insights, there are two 1 % cutoffs.
Advanced Filtering: The first 1 % cutoff essentially removes many false-positive variants that are wrongfully called by GATK. Remember, GATK by design ‘overcalls’ variants because it is not a typical somatic variant caller.
In a typical run, there are 80 – 90 % low-frequency variants at ~0.1 – 0.5 % population frequency that are likely false positives. The cutoff effectively discards those low-quality data points. However, that does not mean that there might be variants present below 1 % population frequency that are true positives and real. If you know what you are looking for, e.g., a relapse-causing variant in a clinical remission sample, this threshold may be lowered, and variants can be identified specifically and selectively. However, thresholds < 1 % cannot be used for de novo low-freq variant discovery.
Review Variants: The second 1 % cutoff removes potential allele dropout (ADO) clones. However, if you select a low-frequency variant that is of high-quality, this clone might end up as a small subclone. In that case, the threshold may be lowered to specifically include the variant-containing clone < 1 %. Similar to advance filtering, this filter removes likely false positives but may introduce a few false negatives.