AdaTiSS

AdaTiSS estimates tissue-specificity (TS) scores from large-scale gene expression data, using data-adaptive robust methods to quantify tissue-specific gene expression across diverse tissues such as those in the Genotype-Tissue Expression (GTEx) project.

Key Features:

• Data-Adaptive Robust Estimation: AdaReg (Adaptive Regression) applies density-power-weighting to robustly estimate population parameters under unknown outlier distributions and non-vanishing proportions of outliers.
• Gaussian-Population Mixture Model: Models expression heterogeneity with a Gaussian-population mixture model to improve estimation of population parameters relevant to tissue specificity.
• AdaTiSS Algorithm: Profiles and standardizes tissue-specificity (TS) scores for each gene across tissues, producing a standardized metric for comparative analysis.

Scientific Applications:

• Tissue-Specific Gene Identification: Enables precise identification of genes with tissue-specific expression patterns across multiple tissues.
• Molecular Tissue Function Analysis: Supports analysis of molecular functions at the tissue level by quantifying TS for genes.
• Disease Mechanism and Drug Targeting Studies: Informs studies of disease mechanisms, drug targeting, and personalized medicine by highlighting tissue-specific gene expression.
• Population-Level Expression Analysis: Facilitates population-level insights into heterogeneous gene expression from large datasets such as GTEx.

Methodology:

The method defines expression populations, models heterogeneity with a Gaussian-population mixture, applies AdaReg with density-power-weighting for robust parameter estimation under unknown outlier conditions, and constructs standardized TS scores per gene.