Methylated fingerprinting technology can help improve the diagnosis of brain tumors

study of DNA
methylation data could improve the diagnosis of brain tumours, according to a study published online today in the University of New China.
correct diagnosis of tumors is essential for cancer treatment. But of the about 100 known tumors, central nervous system tumors are particularly difficult to identify accurately. To solve this problem, Stefan Pfister of the German Cancer Research Center in Heidelberg and colleagues developed a machine learning program that classifys methylated data. Methylation is the process by which methyl molecules attach to DNA to change the availability of information in DNA. This process occurs not only naturally during the normal functioning of the cells - izing each cell with a unique methylated fingerprint - but also in diseases such as cancer. Therefore, methylation can reveal information about tumor types and what types of cells these tumors are formed by.
developed by the authors of the study, 91 tumors can be identified using methylated fingerprints. The training was based on reference data from about 2,800 cancer patients. The authors tested 1,104 artificially examined central nervous system tumors and found that 12 percent were misdiagnosed. Not only does the procedure improve diagnostic accuracy, but its objectivity also allows it to truthfully identify new rare tumors -- artificial examinations that put pressure on diagnosis based on known tumor types, even in atypical cases.
to make this new method widely available, the authors have built a free online tool that analyzes uploaded data in minutes. Since it went online in December 2016, the tool has been used more than 4,500 times, sharing their data to further optimize the algorithm. The authors conclude that integrating methylated fingerprints with an automatic brain tumor classifier could also provide a blueprint for creating similar tumor classification algorithms to diagnose other cancer types. (Source: Feng Weiwei, China Science Journal)