LAUSR

Patient derived xenografts (PDX) are increasingly utilized in translational research and drug development. Characterizing the genomic features of PDX is essential to establishing reliable models for cancer research. Despite great interest, problems remain in PDX tumor data banks including improper cancer type diagnosis and sample mix-ups. In an effort to improve annotation and quality of PDX models, we developed a machine learning model trained on gene expression data from The Cancer Genome Atlas (TCGA). We then applied the model to corresponding data collected from nearly 300 Certis PDX models plus publicly available data from NCI’s Patient-Derived Models Repository (PDMR). The model shows high precision and variable recall and provides a fast and accurate method for cancer type diagnosis. Citation Format: Warren Andrews, Long Do, Jonathan Nakashima. Gene expression based machine learning classifier to predict and validate cancer type in patient derived xenograft (PDX) models. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4299.