AIVariant: a deep learning-based somatic variant detector for highly contaminated tumor samples

• Published in: Experimental & molecular medicine 2023, 55(0), , pp.1734-1742
• Main Authors: Jeon, Hyeonseong, Ahn, Junhak, Na, Byunggook, Hong, Soona, Sael, Lee, Kim, Sun, Yoon, Sungroh, Baek, Daehyun
• Format: Article
• Language: English
• Subjects: 45/23; 631/114/1305; 631/114/1314; 692/699/67; Artificial intelligence; Biomedical and Life Sciences; Biomedicine; Cancer; Cancer research; Deep learning; Genetic diversity; Medical Biochemistry; Molecular Medicine; Nucleotide sequence; Stem Cells; Tumor cells; Tumors; 생화학
• ISSN: 2092-6413; 1226-3613; 2092-6413
• DOI: 10.1038/s12276-023-01049-2

The detection of somatic DNA variants in tumor samples with low tumor purity or sequencing depth remains a daunting challenge despite numerous attempts to address this problem. In this study, we constructed a substantially extended set of actual positive variants originating from a wide range of tumor purities and sequencing depths, as well as actual negative variants derived from sequencer-specific sequencing errors. A deep learning model named AIVariant, trained on this extended dataset, outperforms previously reported methods when tested under various tumor purities and sequencing depths, especially low tumor purity and sequencing depth. Cancer: Detecting genetic variants with AI An artificial intelligence (AI) system trained to detect genetic variants in tumor cell DNA sequence data out-performs existing methods, especially for impure samples contaminated with healthy cells. Identifying the DNA sequence variants in cancer cells is essential for understanding the genetic basis of specific cancers. One major obstacle has been the low proportion of tumor cells in many clinical samples and the challenge of obtaining sequence data specific to the cancer cells. This is especially problematic with some common types of cancer, including those affecting the lungs, kidneys, head, neck and pancreas. Researchers in South Korea led by Daehyun Baek at Seoul National University developed the ‘AIVariant’ method to address these difficulties. The high precision of AIVariant analysis significantly advances the state-of-the-art, which should benefit cancer research and diagnosis, and the search for better treatments.