TIVAN: tissue-specific cis-eQTL single nucleotide variant annotation and prediction.

Area of Science:

• Genomics
• Computational Biology
• Genetic Epidemiology

Background:

• Predicting genetic regulatory variants in non-coding regions is a significant challenge.
• Cis-eQTL single nucleotide variants (SNVs) are crucial for regulating gene expression and understanding phenotypes.
• Existing computational methods for cis-eQTL SNV prediction are underdeveloped compared to pathogenicity prediction.

Purpose of the Study:

• To develop and validate an accurate computational method for predicting tissue-specific cis-eQTL SNVs.
• To provide a tool that annotates and predicts cis-eQTL SNVs in a context-specific manner.
• To improve the characterization of genetic variant impacts on gene expression across different tissues.

Main Methods:

• Developed TIVAN (TIssue-specific Variant ANnotation and prediction), an ensemble method utilizing decision trees.
• Trained TIVAN on a comprehensive dataset incorporating genome-wide genomic and epigenomic profiling data.
• Evaluated performance using five-fold cross-validation (CV-AUC) and Leave-One-Chromosome-Out (LOCO-AUC) metrics across 44 tissues.

Main Results:

• TIVAN accurately discriminates cis-eQTL SNVs from non-eQTL SNVs.
• Achieved higher CV-AUC and LOCO-AUC values compared to other methods across 27 tissue classes.
• Demonstrated consistent top performance on an independent test dataset comprising 7 tissues from 11 studies.

Conclusions:

• TIVAN is a robust and accurate tool for predicting tissue-specific cis-eQTL SNVs.
• The method effectively leverages multi-omics data for variant annotation.
• TIVAN offers a valuable resource for genetic research requiring context-specific variant impact analysis.