Needlestack: an ultra-sensitive variant caller for multi-sample next generation sequencing data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Next-generation sequencing (NGS) enables comprehensive DNA variation analysis.
• Detecting low-abundance somatic mutations (e.g., subclonal, circulating tumor DNA) remains challenging.
• Distinguishing true mutations from sequencing artefacts is difficult at low variant allele frequencies.

Purpose of the Study:

• To develop a highly sensitive variant caller, Needlestack, for accurate somatic mutation detection.
• To address the challenge of distinguishing rare mutations from sequencing errors.
• To improve variant calling performance, especially for low-abundance mutations.

Main Methods:

• Needlestack employs a novel approach that dynamically estimates sequencing error rates by analyzing multiple samples together.
• The method learns the level of systematic sequencing errors directly from the data.
• Performance is evaluated across various mutation types and compared to state-of-the-art methods.

Main Results:

• Needlestack accurately calls mutations by learning data-driven sequencing error profiles.
• The sequencing error rate is shown to vary across different types of alterations, necessitating precise estimation.
• Needlestack demonstrates robustness and outperforms existing methods for detecting low-abundance somatic mutations.

Conclusions:

• Needlestack provides a robust and sensitive solution for identifying low-abundance somatic mutations.
• The dynamic estimation of sequencing error rates is crucial for accurate variant calling.
• Needlestack is freely available, facilitating its adoption in genomic research.