Translating transcriptomic findings from cancer model systems to humans through joint dimension reduction

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• Model systems are crucial for cancer research but may not accurately reflect human biology.
• Inaccuracies in model systems can lead to inconclusive experiments and misleading results.
• There is a need for improved methods to translate findings from model systems to human-relevant data.

Purpose of the Study:

• To present a novel process for applying joint dimension reduction (jDR) to integrate gene expression data.
• To horizontally integrate data across diverse model systems and human tumor cohorts.
• To enhance the translation of model system findings into human-relevant data.

Main Methods:

• Applied joint dimension reduction (jDR) to horizontally integrate gene expression data.
• Combined human TCGA (The Cancer Genome Atlas) gene expression data with data from human cancer cell lines.
• Integrated data from mouse model tumors with human and cell line data.

Main Results:

• Successfully identified aspects of genomic variation that jointly act across different cohorts.
• Demonstrated improvement in predictive modeling derived from model systems.
• Showcased enhancement of clinical biomarkers originating from model systems.

Conclusions:

• The presented jDR approach effectively integrates multi-cohort gene expression data.
• This method improves the human relevance of findings from cancer model systems.
• jDR offers a pathway to more accurate predictive modeling and biomarker discovery in cancer research.