Advantages
- Unaffected by errors due to differences by modalities, equipment, and datasets.
- Utilize comprehensive data rather than limiting to specific biomarkers.
- Can be applied to various diseases other than cancer.
- Applicable in multi-omics analysis.
Background and Technology
There are many studies exploring specific biomarkers for cancer detection or cancer subtype prediction, but due to the heterogeneity of cancer patients, the accuracy of each biomarker is limited. Therefore, utilizing comprehensive biomolecular data is important, however, NGS/microarray data obtained from different batches is difficult to merge due to the batch-to-batch errors (variations by devices, facilities, and timings). k-TSP is one of the common approaches to this error. Comparisons of some pairs of gene expressions (“which is larger”) are used in k-TSP, which means that k-TSP is designed to absorb the error by making the information fuzzy and sacrificing the accuracy.
The present invention is an analysis method and machine learning model that makes full use of comprehensive NGS/microarray data and enables comparison and merging of analysis results from different methods, devices, and datasets. The invention enables cancer detection/cancer subtype prediction using NGS/microarray data from a single sample. This method is also applicable in multi-omics analysis.
 |
Reference
- AACR Special Conference in Cancer Research: Pancreatic Cancer (2023, Boston)
Principal Investigator
Taisuke BABA (Nagoya University Hospital)
Current Stage and Next Step
- There is a big difference between NGS and microarray data due to their difference in modalities. We successfully merged NGS and microarray data using this invention and accurately predicted Bailey’s molecular subtype of PDAC (Figure).
- We are looking for collaborative partners such as transcriptome companies, RNA-Seq/Microarray device manufacturers, diagnostics, clinical testing centers, and software developers to proceed with program development, dedicated device development, and cost verification.
- We aim to apply it to early cancer detection by miRNA profile through health checkups and subtype prediction for personalized medicine.
Project No.BK-04674