Prior knowledge transfer across transcriptional data sets and technologies using compositional statistics yields new...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Transcriptional profiling generates gene lists for patient stratification and biomarker assessment.
• Archived public datasets are crucial for in silico validation but pose data integration challenges.

Purpose of the Study:

• Introduce gene expression compositional assignment (GECA) for validating prior knowledge transfer into independent datasets.
• Demonstrate GECA's ability to overcome data integration issues and outperform existing methods.

Main Methods:

• GECA utilizes compositional statistics for knowledge transfer validation.
• The method does not require normalization of expression levels between datasets.
• GECA was validated using cross-platform gene list transfer in various cancer domains.

Main Results:

• GECA successfully transferred gene lists across platforms for bladder cancer staging and tumor origin prediction.
• Effectively transferred an epithelial ovarian cancer prognostic gene signature from microarrays to next-generation sequencing.
• Identified the OVCAR-5 cell line as gastrointestinal, not ovarian, in origin.

Conclusions:

• GECA is a powerful and simple method for validating prior knowledge transfer in independent datasets.
• It offers an alternative to normalization-dependent methods and improves cross-platform data integration.
• GECA has broad applications in cancer research, biomarker validation, and cell line authentication.