EAGLE: Explicit Alternative Genome Likelihood Evaluator.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Detecting genome variations, particularly insertions and deletions (indels), from single-sample DNA sequencing is challenging due to alignment uncertainties.
• Existing quality filtering methods often yield false positives, necessitating better ways to evaluate variant support.
• Limited options exist for rigorously assessing the data support for candidate variants, hindering validation prioritization.

Purpose of the Study:

• To present EAGLE, a novel method for evaluating the data support for candidate genome variants.
• To provide a robust approach for assessing the probability of observed sequencing data under explicit genomic hypotheses.
• To improve the reliability and specificity of variant calling in DNA sequencing data.

Main Methods:

• EAGLE formulates explicit hypotheses about an individual's genome incorporating candidate variants.
• It computes the probability of observed sequencing data (reads) under each hypothesis.
• The method accounts for alignment uncertainties, multi-mapping, and local misalignments, utilizing the full read length.

Main Results:

• EAGLE demonstrated marked improvement in ranking true insertion/deletion variants on simulated and whole-genome sequencing data.
• A modest but statistically significant improvement was observed on exome sequencing data.
• EAGLE consistently ranked true variants higher than scores from established variant callers.

Conclusions:

• EAGLE enhances the specificity of variant calling by improving the ranking of true variants.
• The method provides a robust way to evaluate data support for candidate genome variations.
• EAGLE is freely available, offering a valuable tool for genomic research and validation.