Genome interpretation using in silico predictors of variant impact

Area of Science:

• Genomics
• Bioinformatics
• Personalized Medicine

Background:

• Clinical associations and lab experiments can only analyze a small fraction of genetic variants.
• The majority of variants are classified as variants of unknown significance (VUS).
• In silico methods offer large-scale variant effect predictions, bridging the gap in genomic data interpretation.

Purpose of the Study:

• To review the training and validation of in silico variant effect prediction methods.
• To illustrate experimental and clinical applications of these predictive tools.
• To assess the performance, strengths, and limitations of various in silico prediction methods.

Main Methods:

• Review of in silico variant effect prediction methodologies.
• Analysis of training and validation datasets for predictive models.
• Comparative assessment of different prediction tools using unpublished datasets.

Main Results:

• In silico methods provide rapid, large-scale estimates of variant effects, primarily using cross-species genetic differences.
• Despite concerns about individual reliability, these methods show practical utility in cohort-based genome interpretation.
• Objective performance evaluations reveal distinct strengths and limitations for each prediction method.

Conclusions:

• In silico variant impact predictors are valuable tools for genome interpretation when applied appropriately.
• Understanding the strengths and limitations of these methods is essential for their effective use.
• Cautious application of in silico tools is necessary to address challenges in interpreting genetic variants.