Comparison of Open-access Databases for Clinical Variant Interpretation in Cancer: A Case Study of MDS/AML

BACKGROUND: Recently, guidelines for variant interpretation in cancer have been established. However, these guidelines do not mention which databases are most suited to performing this task. MATERIALS AND METHODS: We give an overview of existing databases and evaluate their benefit in practical application. We compared three meta-databases and 12 databases for a dataset of patients with myelodysplastic syndrome or acute myeloid leukemia. RESULTS: Clinical implications were found for 13% of all variants. One-third of variants with therapeutic implications were uniquely contained in one database. The VICC meta-database was the most extensive source of information, featuring 92% of variants with a drug association. However, a comparison of meta-databases and original sources indicated that some variants are missing in those meta-databases. CONCLUSION: Public databases provide decision support for interpreting variants but there is still need for manual curation. Meta-databases facilitate the use of multiple resources but should be interpreted with caution.