Shambhala

Shambhala harmonizes human gene expression datasets from microarray hybridization and RNA sequencing by transforming profiles into a universal format compatible for comparative analyses.

Key Features:

• Multi-Dataset Harmonization: Converts multiple gene expression datasets simultaneously to enable comparative studies across extensive collections of samples.
• Platform-Agnostic Approach: Processes data from different experimental platforms, including microarray hybridization and RNA sequencing, without introducing platform-specific bias.
• Calibration Using Auxiliary Datasets: Calibrates each gene expression profile using an auxiliary standardization dataset and transforms profiles to resemble output from the Affymetrix Human Gene microarray hybridization platform.
• Retention of Biological Features: Demonstrates superior retention of biologically important, sample-specific features compared to methods such as quantile normalization and DESeq/DESeq2 normalization.

Scientific Applications:

• Cross-platform integration: Enables integration of gene expression data from microarray and RNA-seq platforms for comparative analyses and meta-analyses.
• Hierarchical clustering validation: Supports biologically meaningful hierarchical clustering of samples independently profiled across three and four different experimental platforms.
• Large-scale comparative studies: Facilitates large-scale comparative studies that require preservation of biological signals across diverse datasets.

Methodology:

Calibration of gene expression profiles using an auxiliary standardization dataset and transformation of each profile to mimic the Affymetrix Human Gene microarray hybridization output; harmonized data were subjected to hierarchical clustering.