Improving the accuracy of NGS-based precision diagnostic results and securing product competitiveness.

NGeneBio announced on last July 3rd, has published their research on gene-specific machine learning for pathogenicity prediction of rare BRCA1 and BRCA2 missense variants in the renowned scientific journal "Scientific Reports" by Nature.

This research paper is about compared to the disease-specific approach, gene-specific supervised machine learning is even more specific as it builds pathogenicity predictors using variants from only a particular disease gene, e.g., BRCA1 or BRCA2. This method has the potential to perform best due to its highest specificity.

Currently, the determination of the pathogenicity of genetic variants relies on information recorded in public databases, which only contain information on a small fraction of frequently occurring variants.  NGeneBio has stated that the main focus of their research is to use machine learning algorithms to assess the pathogenicity of rare variants for which there is insufficient genetic information in these public databases.

In this collaborative study between NGeneBio's research team and Professor Kyu-Baek Hwang's team from Soongsil University, they further investigated the advantages of “gene-specific” machine learning compared to “disease-specific” machine learning. We used 1068 rare (gnomAD minor allele frequency (MAF) < 0.005) missense variants of 28 genes associated with hereditary cancers for our investigation. Popular machine learning classifiers were employed: regularized logistic regression, extreme gradient boosting, random forests, support vector machines, and deep neural networks. As features, we used MAFs from multiple populations, functional prediction and conservation scores, and positions of variants. 

Despite the difficulty in interpreting the association between a specific variant and breast cancer due to insufficient data during the examination of rare genetic variants, they demonstrated the effective interpretation of rare gene mutations using machine learning even with a limited dataset. The most significant achievement was the improvement in prediction and analysis accuracy compared to other approaches by applying it to the prediction of pathogenic mutations in the BRCA1/2 genes associated with breast cancer.

NGeneBio plans to apply these research findings in their breast cancer precision diagnostics product, BRCAaccuTest™PLUS. The accuracy of diagnostic results will be enhanced for breast cancer patients who were previously challenging to determine or diagnose as pathogenic due to lacked sufficient information. Currently, BRCAaccuTest is approved by the MFDS in Korea, being used in more than 10 major domestic hospital including Seoul National University Hospital and Seoul St. Mary's Hospital, as well as overseas medical institutions, such as in Singapore. It is expected that enhancing its competitiveness both domestically and internationally.

Kwang-Joong Kim, the Head of NGeneBio R&D said that "By using gene-specific machine learning, we have been able to identify variants with inaccurate interpretations and clearly distinguish between clinically significant and non-significant variants. This has led to improved accuracy in NGS-based precision diagnostic results." He further added, "We will expand the application of this approach to all gene panel analyses based on NGS, enabling accurate diagnosis of challenging disease-associated genetic variants."

Gene-specific machine learning for pathogenicity prediction of rare BRCA1 and BRCA2 missense variants | Scientific Reports (nature.com)