DTDHM: detection of tandem duplications based on hybrid methods using next-generation sequencing data

Area of Science:

• Genomics and Bioinformatics
• Structural Variation Analysis
• Next-Generation Sequencing (NGS) Data Interpretation

Background:

• Tandem duplications (TDs) are significant structural variations in the human genome.
• TDs are implicated in the pathogenesis of various diseases, including cancer.
• Accurate TD detection is challenging due to uneven read distribution and NGS data complexity.

Purpose of the Study:

• To develop and evaluate a novel method for detecting tandem duplications (TDs) from NGS data.
• To improve the accuracy and reliability of TD detection, particularly in challenging genomic contexts.

Main Methods:

• Proposed DTDHM (detection of tandem duplications based on hybrid methods) pipeline integrating read depth (RD), split read (SR), and paired-end mapping (PEM) signals.
• Employed K-nearest neighbor (KNN) algorithm for multi-feature classification to address uneven sample distribution.
• Validated DTDHM against three other methods using 450 simulated and five real human genome datasets.

Main Results:

• DTDHM achieved the highest average F1-score (80.0%) across 450 simulated datasets, outperforming SVIM, TARDIS, and TIDDIT.
• DTDHM demonstrated superior stability and accuracy, with a 1.2x higher detection effect than the next best method and minimal boundary bias (around 20 bp).
• In real data experiments, DTDHM yielded the highest overlap density score (ODS) and F1-score, confirming its effectiveness.

Conclusions:

• DTDHM offers excellent sensitivity, precision, F1-score, and boundary accuracy for TD detection from NGS data.
• The method proves reliable, especially for samples with low coverage depth and tumor purity.
• DTDHM represents a robust tool for advancing structural variation analysis in human genomics.