Related Experiment Video
Updated: Mar 3, 2026
Detecting Somatic Genetic Alterations in Tumor Specimens by Exon Capture and Massively Parallel Sequencing
Published on: October 18, 2013
Reliability of algorithmic somatic copy number alteration detection from targeted capture data
Nora Rieber1, Regina Bohnert1, Ulrike Ziehm1
1Molecular Health GmbH, Kurfürsten-Anlage 21, 69115 Heidelberg, Germany.
Motivation:
Whole exome and gene panel sequencing are increasingly used for oncological diagnostics. To investigate the accuracy of SCNA detection algorithms on simulated and clinical tumor samples, the precision and sensitivity of four SCNA callers were measured using 50 simulated whole exome and 50 simulated targeted gene panel datasets, and using 119 TCGA tumor samples for which SNP array data were available.
Results:
On synthetic exome and panel data, VarScan2 mostly called false positives, whereas Control-FREEC was precise (>90% correct calls) at the cost of low sensitivity (<40% detected). ONCOCNV was slightly less precise on gene panel data, with similarly low sensitivity. This could be explained by low sensitivity for amplifications and high precision for deletions. Surprisingly, these results were not strongly affected by moderate tumor impurities; only contaminations with more than 60% non-cancerous cells resulted in strongly declining precision and sensitivity. On the 119 clinical samples, both Control-FREEC and CNVkit called 71.8% and 94%, respectively, of the SCNAs found by the SNP arrays, but with a considerable amount of false positives (precision 29% and 4.9%).
Discussion:
Whole exome and targeted gene panel methods by design limit the precision of SCNA callers, making them prone to false positives. SCNA calls cannot easily be integrated in clinical pipelines that use data from targeted capture-based sequencing. If used at all, they need to be cross-validated using orthogonal methods.
Availability And Implementation:
Scripts are provided as supplementary information.
Contact:
gunther.jansen@molecularhealth.com.
Supplementary Information:
Supplementary data are available at Bioinformatics online.
Insights
Evaluating SCNA callers on whole exome and gene panel sequencing data revealed significant false positive rates. Cross-validation with orthogonal methods is crucial for reliable oncological diagnostics.
Area of Science:
- Genomics
- Bioinformatics
- Oncology
Background:
- Whole exome and gene panel sequencing are increasingly utilized in oncological diagnostics.
- Accurate detection of somatic copy number alterations (SCNAs) is critical for cancer research and clinical applications.
Purpose of the Study:
- To evaluate the accuracy, precision, and sensitivity of different SCNA detection algorithms.
- To compare algorithm performance on simulated and clinical tumor samples using whole exome and targeted gene panel sequencing data.
Main Methods:
- Assessed four SCNA callers using simulated whole exome and gene panel datasets (n=50 each).
- Validated performance on 119 clinical TCGA tumor samples with available SNP array data.
- Investigated the impact of tumor purity on SCNA detection accuracy.
Main Results:
- VarScan2 produced numerous false positives on synthetic data. Control-FREEC showed high precision but low sensitivity. ONCOCNV demonstrated similar limitations, particularly for amplifications.
- Tumor purity above 60% significantly decreased precision and sensitivity for all callers.
- On clinical samples, Control-FREEC and CNVkit detected 71.8% and 94% of SNP array-identified SCNAs, respectively, but with substantial false positive rates.
Conclusions:
- Current whole exome and gene panel sequencing methods inherently limit SCNA caller precision, leading to false positives.
- Integrating SCNA calls from targeted capture-based sequencing into clinical pipelines is challenging.
- Orthogonal method cross-validation is essential for reliable SCNA calling in clinical settings.
