Analytical Performance Characteristics
Analytical performance was established for Franklin by QIAGEN using the default exome workflow under validated study conditions. Analytical performance was evaluated for short variant calling and copy number variant calling using reference materials and simulation-based studies. Performance was established for analyses performed against the GRCh37/hg19 and GRCh38/hg38 human reference genomes.
Under the validated study conditions, the device demonstrated acceptable performance for the variant types and workflows specified above. The established analytical performance applies only to the validated software versions, workflow configuration, input types, and study conditions. Performance may differ when the software is used with other workflow settings, data characteristics, third-party upstream processes or laboratory-specific implementation conditions.
The device is a software-based IVD and does not directly measure an analyte. Therefore, the following analytical performance characteristics are not applicable to this device: analytical specificity, trueness (bias), accuracy as a measurement parameter, limit of detection, measurement range, linearity, and reference measurement procedures.
Under validated study conditions, the device demonstrated the following analytical performance characteristics: small variants (SNP and indels) sensitivity ≥ 0.995, small variants (SNP and indels) precision ≥ 0.998, CNV deletion sensitivity ≥0.997, CNV duplication sensitivity ≥ 0.973. Performance applies only to the validated workflow, software version, reference genome, and input data characteristics. Performance may be reduced outside these validated conditions.
Clinical Performance Characteristics
Clinical performance of the device was established for Franklin by QIAGEN using the default exome workflow under validated study conditions. Clinical performance was evaluated in controlled studies covering annotation, filtering, germline analysis, and somatic analysis. These studies demonstrated that the device provides clinically relevant and consistent interpretation support within the validated use conditions.
Clinical performance applies only to the validated software version, workflow configuration, intended input data, and study conditions. Clinical performance may vary depending on workflow configuration, input data quality and characteristics, use environment, and laboratory procedures.
The device supports professional interpretation of genetic variants and does not independently establish a diagnosis or predict a clinical outcome without user review.
Therefore, the following clinical performance characteristics are not applicable to this device: threshold value, diagnostic specificity, positive predictive value, negative predictive value, likelihood ratio, and expected values in normal and affected populations.
Validated annotation performance included gene annotation accuracy of >99%, region classification accuracy of ≥95%, effect classification accuracy of ≥98.5%.
Mathematical Principles and Basis of Analysis
The device generates analytical results using a bioinformatic workflow referenced to the selected human reference genome (GRCh37/hg19 or GRCh38/hg38).
For supported sequence data inputs processed within the validated workflow, the device performs short variant calling for SNPs and indels, and copy number variant calling for deletions and duplications.
For externally generated VCF or other supported variant call files, the device does not recalculate the original variant calls and instead uses the imported variant calls as the basis for downstream annotation, filtering, review and interpretation support.
Analytical performance was established using the following parameters:
Analytical Sensitivity: TP/(TP + FN)
Analytical Precision: TP/(TP + FP)
Where:
True positives (TP): variants which were called by the pipeline and were also present in the reference set.
False negatives (FN): variants not identified by the pipeline but were present in the reference set.
False positives (FP): variants which were called by the pipeline but were absent in the reference set.
Results generated by the device depend on the selected workflow configuration, the reference genome, the characteristics and quality of the input data, and any upstream third-party processing steps.
Limitations
The established analytical and clinical performance applies only to the validated study conditions described for the device. Performance outside the validated conditions has not been established by the manufacturer.
The device is intended to support qualified professional users in the review and interpretation of genomic data. The results provided by the device shall be reviewed by appropriately trained personnel and interpreted in the context of the full clinical and laboratory information available for the case.
Local Validation
Before routine clinical use, the user laboratory shall perform appropriate local verification and validation, as applicable, in its own operating environment and in accordance with applicable standards and internal procedures.
Local validation should confirm, as applicable, that the implemented workflow, input data types, upstream processing steps, configuration settings and reporting practices are suitable for the laboratory’s intended use.