Bioinformatics Seminars

Bioinformatics Seminar

Time: 11AM
Venue: Davis Auditorium and Teams

7 June 2022

Evaluating cancer to model similarity with gene expression profiling

Sam Lee
WEHI Bioinformatics

Relating insights from pre-clinical studies to individual patient tumour samples relies on the selection of the most relevant pre-clinical models. Analysis of transcriptional profiles is a promising approach for identifying the best pre-clinical model matches for primary tumours, however benchmarking of existing methods has been limited. To address this, we have developed a benchmarking strategy that relies on the assessment of sample similarity both within and across model types to test a wide variety of standard and bespoke methods. Using comparisons of cell-lines, patient derived xenografts, and primary tumour samples, we find that most published methods do not outperform a simple, non-parametric approach for selecting similar models. This highlights the importance of carefully selecting an analysis approach that will enable confident selection of pre-clinical models with the most relevance for patient tumour samples.

Search past seminars