SPECS

SPECS identifies tissue-specific molecular features using a non-parametric specificity score compatible with unequal sample group sizes and evaluates protein structural models by assessing structural similarity with a united-residue model that integrates side-chain orientation and global distance-based measures.

Key Features:

• Tissue-specificity scoring: Implements a non-parametric specificity score that supports unequal sample group sizes to detect genes with tissue-predominant expression.
• GTEx application: Applied to all GTEx samples to identify known and novel tissue-specific genes.
• Structural similarity metric: Introduces a similarity metric that incorporates side-chain orientation alongside global distance-based measures for model versus native comparisons.
• United-residue model: Uses a united-residue model of polypeptide conformation to represent side-chain orientation within structural assessments.
• Side-chain sensitivity: Captures minute variations in side-chain χ angles and detects local conformational differences even when Cα traces are accurate.
• Resolution range: Provides reliable evaluation for both high- and moderate-resolution protein models and outperforms metrics based solely on Cα positioning.

Scientific Applications:

• Tissue-specific gene discovery: Identification of genes with tissue-specific or tissue-predominant expression patterns in transcriptomic studies such as GTEx.
• Protein model validation: Quantitative evaluation of predicted protein structures against experimentally determined native structures.
• Benchmarking structure prediction methods: Comparative assessment of modeling approaches using metrics sensitive to side-chain orientation and global distance.
• Local conformational analysis: Detection and characterization of side-chain χ angle deviations and local structural inaccuracies in models.

Methodology:

Computational methods include a non-parametric specificity score for unequal sample group sizes applied to GTEx data, and a united-residue polypeptide conformation model with a similarity metric that integrates side-chain orientation and global distance-based measures to evaluate models against experimentally determined native structures and side-chain χ angles.