FIRE: functional inference of genetic variants that regulate gene expression.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Interpreting genetic variation in noncoding genomic regions is crucial for personal genome analysis.
• Noncoding single nucleotide variants (SNVs) can influence phenotypes by regulating gene expression.
• Predictive methods for SNVs are needed to interpret variants of unknown significance from whole-genome sequencing.

Purpose of the Study:

• To develop a computational tool, FIRE (Functional Inference of Regulators of Expression), for scoring SNVs based on their potential to regulate gene expression.
• To aid in the interpretation of genetic variants identified in large-scale sequencing studies.

Main Methods:

• Developed FIRE, a tool utilizing 23 random forests trained on genomic annotations.
• Trained models to recognize SNVs within cis-expression quantitative trait loci (cis-eQTLs).
• Utilized 92 genomic annotations as predictive features for SNV function.

Main Results:

• FIRE scores effectively discriminate cis-eQTL SNVs from non-eQTL SNVs with a cross-validated AUC of 0.807.
• FIRE demonstrated high accuracy in discriminating shared cis-eQTL SNVs across diverse populations (AUC of 0.939).
• FIRE scores showed predictive power for cis-eQTL SNVs across various tissue types.

Conclusions:

• FIRE provides a valuable tool for assessing the regulatory potential of noncoding and coding SNVs.
• The tool aids in the functional interpretation of genetic variants, particularly those identified through whole-genome sequencing.
• FIRE scores are available for genome-wide SNVs, facilitating broader research applications.