Identification of single nucleotide variants using position-specific error estimation in deep sequencing data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Targeted deep sequencing is crucial for identifying single nucleotide variants (SNVs) in translational medicine and cancer profiling.
• Detecting SNVs, especially at low variant allele frequencies (VAFs), is computationally challenging due to sequencing artifacts and noise.
• Existing methods struggle with analytical sensitivity in identifying low-frequency SNVs.

Purpose of the Study:

• To develop a bioinformatics tool, AmpliSolve, for accurate SNV detection in targeted deep sequencing data.
• To address the challenge of high noise levels in amplicon-based sequencing data.
• To improve the sensitivity and precision of SNV calling, particularly for low VAFs.

Main Methods:

• Developed AmpliSolve, a bioinformatics tool utilizing normal samples to model background noise.
• Implemented a Poisson model-based statistical framework for SNV detection.
• Tested AmpliSolve on synthetic and real sequencing data, including amplicon-based libraries.

Main Results:

• AmpliSolve demonstrates a favorable balance between precision and sensitivity for SNV detection.
• The tool effectively identifies SNVs at VAFs below 5% and down to 1%.
• Validation using digital droplet PCR confirmed AmpliSolve's accuracy in detecting SNVs in circulating tumor DNA samples.

Conclusions:

• AmpliSolve is an effective tool for in-silico estimation of background noise and low-frequency SNV detection in targeted deep sequencing.
• The tool is specifically designed for Ion Torrent amplicon-based libraries but is adaptable to other platforms.
• AmpliSolve is freely available, facilitating its use in various research applications.