P.E.P.P.E.R. and DeepVariant

P.E.P.P.E.R. and DeepVariant perform accurate variant calling and assembly polishing for long-read sequencing data from Oxford Nanopore Technologies (ONT) and Pacific Biosciences (PacBio), enabling haplotype-aware SNV detection and diploid assembly improvement.


Key Features:

  • Haplotype-Aware Variant Calling: Implements a haplotype-aware approach to improve identification of single nucleotide variants (SNVs) in segmental duplications and low-mappability regions.
  • Long-Read Sequencing Integration: Optimized for ONT and PacBio long reads to link adjacent variants and enable read-based phasing, producing contiguous phase blocks spanning 85%–92% of annotated genes.
  • Superior Performance with Nanopore Data: Addresses pore-based signal error profiles to achieve higher whole-genome SNV identification accuracy on nanopore data compared with short-read methods.
  • Extension to PacBio HiFi Data: Extends to PacBio High-Fidelity (HiFi) reads and reports performance exceeding methods such as WhatsHap-DeepVariant.
  • De Novo Assembly Polishing: Performs assembly polishing with nanopore and PacBio HiFi reads to produce diploid assemblies with reported quality scores of Q35+ for nanopore-polished assemblies and Q40+ for PacBio-HiFi-polished assemblies.

Scientific Applications:

  • Complex-region variant detection: Detects SNVs in segmental duplications and other low-mappability genomic regions using long-read data.
  • Structural variation and haplotype phasing: Enables read-based phasing and generation of contiguous phase blocks for structural variation and haplotype analyses.
  • Diploid de novo assembly improvement: Polishes de novo assemblies with nanopore and PacBio HiFi reads to increase assembly quality values (QV).
  • Population and medical genomics: Supports genome-wide variant discovery applicable to personalized medicine, evolutionary biology, and population genetics studies.

Methodology:

PEPPER uses recurrent neural networks for candidate detection and is integrated into the PEPPER‑Margin‑DeepVariant pipeline with DeepVariant for genotyping, employing read-based phasing and assembly polishing using nanopore and PacBio HiFi reads.

Topics

Details

License:
MIT
Cost:
Free of charge
Tool Type:
workflow
Programming Languages:
Python, C++
Added:
5/27/2021
Last Updated:
8/2/2021

Operations

Publications

Shafin K, Pesout T, Chang P, Nattestad M, Kolesnikov A, Goel S, Baid G, Eizenga JM, Miga KH, Carnevali P, Jain M, Carroll A, Paten B. Haplotype-aware variant calling enables high accuracy in nanopore long-reads using deep neural networks. Unknown Journal. 2021. doi:10.1101/2021.03.04.433952.

Documentation

Downloads

Links