Recently, a collaboration between the research team led by Dr. Weiliang Zhu and Dr. Zhijian Xu from the Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Associate Professor Dong Xie's team from the Shanghai Pulmonary Hospital, and Dr. Ying Ding's team from the First Affiliated Hospital of Nanjing Medical University has resulted in the establishment of a clinical medication database for patients with epidermal growth factor receptor (EGFR) mutations and a predictive model for EGFR mutation drug sensitivity. This development holds the promise of assisting clinical physicians in designing rational and reliable personalized treatment regimens. The related research has been published in the "Briefings in Bioinformatics" journal.
EGFR is one of the most extensively studied driver genes in lung cancer and serves as a crucial target for the development of targeted therapies for non-small cell lung cancer. With the advancement of gene sequencing methods, many clinically significant novel EGFR mutation types have been identified, posing new challenges for individualized precision medicine for patients harboring such mutations.
The research team gathered clinical treatment-related literature for EGFR mutant lung cancer patients over the past two decades and constructed the D3EGFRdb, a clinical medication database for EGFR mutation patients. Leveraging the clinical case analysis results, the research team utilized the DeepPurpose library to construct 80 different combinations of deep learning models and performed screening, further applying 17 models to the clinical dataset of D3EGFRdb. The results indicated that the best comprehensive predictive performance was achieved when the drug encoder was Morgan and the mutation protein encoder was CNN. Consequently, the research team adopted this as the final model, named D3EGFRAI. Building upon this achievement, the team established the D3EGFR online platform, which is freely accessible to all users. The platform offers users case retrieval for EGFR mutant patients and clinical drug response prediction services.
The drug recommendation platform for EGFR-mutant lung cancer, D3EGFR, is featured in the "Briefings in Bioinformatics" journal.
Related paper information: https://doi.org/10.1093/bib/bbae121
