iFACT QC Threshold
The individualized Fetal Aneuploidy Confidence Test (iFACT) is a dynamic threshold QC metric that is assessed for each sample. This metric indicates whether the system has generated sufficient sequencing coverage for the FF estimate to make a call for each sample. If a sample fails iFACT QC, the system does not generate a result.
The iFACT QC consists of the following components:
| 1. | Samples with insufficient coverage (ie, below a fixed coverage threshold) are flagged to be repeated from plasma, regardless of their FF estimate. |
| 2. | Samples are compared with a dynamic threshold (eg, the iFACT QC curve) that considers the relationship between coverage and FF. |
Samples with low coverage can pass this threshold if they have sufficient FF, while samples with low FF can pass this threshold with sufficient coverage. Samples classified as positive for an aneuploidy never fail QC based on the iFACT QC curve alone because this metric is optimized to favor sensitivity.
The iFACT QC curve was determined from characteristics such as sequencing depth (coverage) and FF learned from clinical samples processed by the Illumina CLIA laboratory. Any estimate of FF contains a certain amount of error relative to the actual underlying true FF of a sample. The distribution of underlying true FF was determined by subtracting the known error in the FF estimates from the observed FF distribution in the clinical samples. These calculations indicate the relationship between observed FF and true FF across the entire range of FF values.
The amount of error in the FF estimate depends on the coverage of the sample. The lower the coverage, the higher the error in the FF estimate. The higher the coverage, the lower the error in the FF estimate. Therefore, the relationship between observed and true FF is also dependent on coverage. Similarly, the LOD95 (level of FF at which 95% sensitivity is achieved) depends on coverage. All this information was used to establish a relationship that determines whether the underlying true FF of a sample is likely to exceed the corresponding LOD95. The resulting iFACT QC curve provides the threshold indicating confidence that the underlying true FF exceeds the LOD95.
