HIGH-PERFORMANCE COMPUTING ENHANCES TREATMENT PRECISION IN BREAST CANCER
The proposed experiment aims to improve personalized drug response prediction tools by analysing cancer patients’ molecular profiles. This intelligent diagnostic platform will identify drugs with high effectiveness, reducing human and economic loss due to genetic and molecular variations.
Start date: 01/06/2021
Duration in months: 18
Problem Description
Breast cancer is a global health issue with 2.3 million cases and 685000 deaths. Treatment failures are attributed to genetic and molecular variations. New genomic technology offers treatment regimens assessing tumour DNA, RNA, protein, and metabolites, requiring large data sets and HPC resources.
Goals
New services
Challenges
CHOSA aims to develop an intelligent platform that can identify drugs most effective for each patient based on their molecular profiles, such as biopsy results, tumour genome, and RNA information. Current technologies only predict treatment outcomes in highly selected cases.
Innovation results
The experiment used HPC to analyse NCI-60 data, linking 60 cancer cell lines to over 50000 compounds. Machine learning models were built for 5986 compounds, with 119 showing significant predictive power. Eight models were interesting for breast cancer, proving the HPC-backed ML approach's value.
Business impact
CHOSA is targeting a €69 million market in Germany and Nordic countries, aiming to increase turnover by several million Euros by mid-2024. This strategy aims to provide lifesaving treatment to limited cancer patients and prolong the lives of advanced cancer patients.