Pivotal role of biallelic frequency analysis in identifying copy number alterations using genome-wide methods in tumors with a high level of aneuploidy.

Chromosome number abnormalities is one of the hallmarks of cancer. DNA copy number alterations (CNA) are studied using various genome-wide methods. In our study, we investigated CNA in human pituitary tumors using three platforms CytoSNP-850 K microarrays, low-pass whole-genome sequencing (average × 7 coverage, LPWGS), and Infinium Methylation EPIC array. Virtual karyotypes based on each dataset were generated using open-source software packages for each sample. Concordant CNA profiles were found for most of tumor. Surprisingly, substantial discrepancies between results from SNP arrays and LPWGS/EPIC arrays were identified in 20% of tumors, for which discrimination of true karyotype was required. B-allelic frequency data from SNP arrays was crucial to adjust normal ploidy level as ultimately verified with FISH. The discrepancy between virtual karyotypes was more pronounced the more CNAs were found. When CNAs covered around half of genome, the level of normal/diploid copy number was incorrectly set with methods, based solely on signal intensity/read-counts coverage. To conclude, CNA analyses with methods such as LPWGS and methylation arrays in highly aneuploid tumors are prone to a bias from improper normal ploidy level setting. These methods are commonly used therefore, we aimed to aware the scientific community about this underestimated methodological problem.