Home

What is deTiN?

The basis of somatic mutation detection is the ability to distinguish between somatic (and therefore tumor-specific) and germline, or inherited, variants (e.g. SNPs). Typically, this is performed by comparing, sequencing data from tumor DNA and patient-matched germline DNA, taken from healthy tissue. In conjunction with our conventional practice of removing recurrent artifacts through comparison with data from a panel of normal samples, the tumor/normal sequencing data comparison removes patient-specific inherited variants, and helps remove alignment and other locus-specific artifacts that affect both samples. This variant detection paradigm allows for sensitive (generally at the level of >95% for SSNVs) and specific somatic mutation calling, resulting in a low false positive rate (<0.5 mutations per megabase).

While such variant detection is both sensitive and specific, the methods rely on the ability to provide sequencing data from matched healthy tissue free of contaminating tumor cells. With a significant contamination of malignant cells in the matched germline control, or tumor-in-normal (TiN) contamination, state-of-the-art detection methods tend to reject true somatic variants based on the presence of tumor-derived reads supporting the mutation in the matched healthy tissue. Thus, TiN (the proportion of tumor DNA found in the normal) can dramatically decrease the sensitivity for mutation detection (i.e. increase false negative rate), adversely affecting the utility of cancer sequencing data for research and clinical use.

Procuring matched normal tissue can be challenging; at times, the only available tissue may be contaminated with tumor cells. TiN contamination can result from sampling techniques such as the use of adjacent normal tissue or saliva (that, for example, can be invaded by leukemic cells). Studies of hematologic cancers have reported substantial TiN even when skin biopsies or saliva were used as matched controls. Several other studies have reported shared somatic loss-of-heterozygosity events in tumors and adjacent normal tissue in breast, bladder and gastric cancers, indicating possible field effects and/or TiN. Furthermore, even attempts to reduce contamination using flow cytometry or histological assessment can fail to provide completely pure normal samples. To overcome the challenges of TiN contamination and reduce false negatives, we developed deTiN, a method that estimates TiN and salvages somatic mutations that would otherwise filtered out, or suspected as germline or artifactual events. By estimating and correcting for TiN, deTiN provides a useful quality control metric and allows for improved analysis of contaminated data.