| Autor: |
Chang LC; Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, MD, USA., Das B; Molecular Characterization and Clinical Assay Development Laboratory, Leidos Biomedical Research Inc., Frederick National Laboratory for Cancer Research, Frederick, MD, USA., Lih CJ; Molecular Characterization and Clinical Assay Development Laboratory, Leidos Biomedical Research Inc., Frederick National Laboratory for Cancer Research, Frederick, MD, USA., Si H; Molecular Characterization and Clinical Assay Development Laboratory, Leidos Biomedical Research Inc., Frederick National Laboratory for Cancer Research, Frederick, MD, USA., Camalier CE; Molecular Characterization and Clinical Assay Development Laboratory, Leidos Biomedical Research Inc., Frederick National Laboratory for Cancer Research, Frederick, MD, USA., McGregor PM 3rd; Molecular Characterization and Clinical Assay Development Laboratory, Leidos Biomedical Research Inc., Frederick National Laboratory for Cancer Research, Frederick, MD, USA., Polley E; Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, MD, USA. |
| Abstrakt: |
With rapid advances in DNA sequencing technologies, whole exome sequencing (WES) has become a popular approach for detecting somatic mutations in oncology studies. The initial intent of WES was to characterize single nucleotide variants, but it was observed that the number of sequencing reads that mapped to a genomic region correlated with the DNA copy number variants (CNVs). We propose a method RefCNV that uses a reference set to estimate the distribution of the coverage for each exon. The construction of the reference set includes an evaluation of the sources of variability in the coverage distribution. We observed that the processing steps had an impact on the coverage distribution. For each exon, we compared the observed coverage with the expected normal coverage. Thresholds for determining CNVs were selected to control the false-positive error rate. RefCNV prediction correlated significantly (r = 0.96-0.86) with CNV measured by digital polymerase chain reaction for MET (7q31), EGFR (7p12), or ERBB2 (17q12) in 13 tumor cell lines. The genome-wide CNV analysis showed a good overall correlation (Spearman's coefficient = 0.82) between RefCNV estimation and publicly available CNV data in Cancer Cell Line Encyclopedia. RefCNV also showed better performance than three other CNV estimation methods in genome-wide CNV analysis. |
