NextSV

NextSV integrates multiple aligners and structural variant callers to detect deletions, insertions, duplications, and inversions from low-coverage Pacific Biosciences (PacBio) long-read human whole-genome sequencing data, producing sensitive and stringent call sets to balance recall and precision.


Key Features:

  • Meta-caller integration: Integrates three aligners and three distinct SV callers into a single meta-caller framework for SV detection.
  • SV type support: Detects deletions, insertions, duplications, and inversions.
  • Dual call sets: Produces a sensitive call set aimed at maximizing recall and a stringent call set focused on enhancing precision.
  • Low-coverage optimization: Optimized for low-coverage PacBio long-read sequencing data and evaluated around ~10X coverage.
  • Benchmarking metrics: Reports precision, recall, and F1 score for comparative performance assessment.
  • Dataset evaluation: Benchmarked using NA12878 and HX1 genomes across varying PacBio coverages.
  • Mendelian error analysis: Assesses Mendelian errors using an Ashkenazi Jewish trio to evaluate inherited SV detection.
  • Coverage recommendation: Indicates that approximately 10X PacBio coverage provides a balance between sequencing cost and effective SV detection.

Scientific Applications:

  • Human structural variant discovery: Detection of deletions, insertions, duplications, and inversions from low-coverage PacBio whole-genome data.
  • SV caller and aligner comparison: Comparative evaluation of multiple aligners and SV callers via meta-calling to improve call reliability.
  • Mendelian inheritance studies: Analysis of inherited structural variants and Mendelian error rates in pedigree datasets such as the Ashkenazi Jewish trio.
  • Sequencing design guidance: Informing experimental design and coverage decisions for long-read sequencing projects targeting SV discovery.

Methodology:

Integrates three aligners and three SV callers into a meta-caller workflow, generates sensitive and stringent call sets, benchmarks performance on NA12878 and HX1 under varying PacBio coverages (reporting deletion recall ~93.5–94.1% and insertion recall ~87.9–93.2% near 10X and F1-based precision/accuracy), and assesses Mendelian errors using an Ashkenazi Jewish trio.

Topics

Details

License:
MIT
Tool Type:
command-line tool, library
Operating Systems:
Linux, Windows, Mac
Programming Languages:
Python
Added:
7/31/2018
Last Updated:
11/25/2024

Operations

Publications

Fang L, Hu J, Wang D, Wang K. NextSV: a meta-caller for structural variants from low-coverage long-read sequencing data. BMC Bioinformatics. 2018;19(1). doi:10.1186/s12859-018-2207-1. PMID:29792160. PMCID:PMC5966861.

Documentation