An RNA seq-based reference landscape of human normal and neoplastic brain

Area of Science:

• Neuroscience
• Oncology
• Bioinformatics

Background:

• Understanding the molecular differences between normal brain tissue and neoplastic conditions is crucial for developing targeted therapies.
• Existing large-scale datasets for brain tumors and normal brain are often fragmented, hindering comprehensive analysis.

Approach:

• Integrated five large-scale public RNA-sequencing datasets, correcting for batch effects.
• Applied Uniform Manifold Approximation and Projection (UMAP) to generate a unified reference landscape (Brain-UMAP) of normal brain and various tumor types.
• Developed an interactive online tool, Oncoscape, for visualizing multi-omic data within this reference landscape.

Key Points:

• The Brain-UMAP reference includes 702 adult gliomas, 802 pediatric brain tumors, and 1409 normal brain samples.
• Distinct clusters emerged for normal brain regions and different tumor types, facilitating comparative analysis.
• Comparative gene expression profiles and gene ontology patterns are readily accessible due to the RNA-sequencing basis of the landscape.

Conclusions:

• This meta-analysis provides a novel resource for comparing gene expression and pathways across adult gliomas, pediatric brain tumors, and normal brain.
• The open-source Oncoscape platform enables visualization of clinical metadata, gene expression, mutations, and copy number alterations.
• This integrated approach enhances the investigation of brain tumor biology and its relationship to normal brain function.