trace

trace predicts tissue-selective causal mechanisms and prioritizes candidate disease genes for Mendelian and rare diseases by applying machine learning to large-scale tissue-aware human gene data.

Key Features:

• Tissue-aware analysis: Leverages heterogeneous, large-scale tissue-aware datasets of human genes for modeling.
• Multi-mechanism assessment: Quantitatively assesses hundreds of candidate mechanisms per disease concurrently.
• Operational modes: Operates in gene-specific, tissue-specific, or patient-specific modes.
• Interpretability: Produces interpretable outputs that clarify how specific genes contribute to tissue-specific disease manifestation.
• Mechanism identification: Applied to selected Mendelian disease genes to pinpoint mechanisms leading to tissue-specific manifestations.
• Cross-disease factor discovery: Reveals known and underappreciated factors underlying tissue-selective manifestation when applied to diseases manifesting in the same tissue.
• Diagnostic prioritization: Creates patient-specific models that prioritized the pathogenic gene in 86% of 50 patients.

Scientific Applications:

• Mechanistic interpretation: Dissects tissue-selective modes of action of Mendelian and rare disease genes.
• Genetic diagnosis: Prioritizes candidate disease-causing genes in patients using tissue-specific models.
• Comparative tissue analysis: Identifies shared and distinct contributing factors across diseases affecting the same tissue.
• Candidate filtering: Filters out unlikely candidate genes by incorporating tissue selectivity into gene prioritization.

Methodology:

Machine learning models trained on heterogeneous, large-scale tissue-aware human gene datasets quantitatively assess hundreds of candidate mechanisms per disease and can be applied in gene-, tissue-, or patient-specific modes.