Advanced search
Add to list

Scheduled maintenance: Publication downloads temporarily unavailable.

Due to maintenance publication downloads will not be available on:

  • Wednesday, March 27, 17:00 – 21:00
  • Thursday, March 28, 17:00 – 21:00

Exports of lists, FWO and BOF information will remain available.

For any questions, please contact biblio@ugent.be. Apologies for any inconveniences, and thank you for your understanding.

Benchmarking of long-read structural variant callers using Oxford Nanopore data

Griet De Clercq (UGent) , Bram Van Gaever (UGent) , Lies Vantomme (UGent) , Annelies Dheedene (UGent) and Björn Menten (UGent)
Author
Organization
Abstract
As long-read sequencing (LRS) technologies mature, several bioinformatics tools designed to identify structural variants (SVs) have been developed. To allow validation of these tools, Zook et al. published a highly curated SV truth set of Genome in a Bottle sample NA24385, consisting of deletions and insertions. We performed a benchmarking analysis with five LRS SV callers against this set utilising in-house generated Oxford Nanopore reads. SVs are called with cuteSV, SVIM, sniffles, pbsv, and nanovar. The callers are assessed in terms of resource usage, reproducibility, and calling performance. The latter is evaluated with Truvari giving recall and precision statistics on SV detection. We further investigate the influence of read support, sequencing coverage, SV type and length, and integration of call sets with Jasmine. CuteSV achieves overall best performance, while nanovar lags behind in both resource usage and calling statistics. A coverage greater than 20x offers no additional advantage for reliable SV detection, while the recommended read support of one third of the coverage proves to be too stringent. Integration of call sets with Jasmine should include three callers to compete with stand-alone call sets. We propose a minimum coverage of at least 15x for optimal sensitivity and specificity. Read support should be set at one fifth of the coverage to obtain optimised calling performance. CuteSV performs best in both sensitivity and specificity, and resource usage. Further work is however needed to assess results for different SV types and more complex regions.

Citation

Please use this url to cite or link to this publication:

MLA
De Clercq, Griet, et al. “Benchmarking of Long-Read Structural Variant Callers Using Oxford Nanopore Data.” Joint BeSHG / NVHG Meeting 2022, Abstracts, 2022.
APA
De Clercq, G., Van Gaever, B., Vantomme, L., Dheedene, A., & Menten, B. (2022). Benchmarking of long-read structural variant callers using Oxford Nanopore data. Joint BeSHG / NVHG Meeting 2022, Abstracts. Presented at the Joint BeSHG/NVHG meeting 2022, Bruges, Belgium.
Chicago author-date
De Clercq, Griet, Bram Van Gaever, Lies Vantomme, Annelies Dheedene, and Björn Menten. 2022. “Benchmarking of Long-Read Structural Variant Callers Using Oxford Nanopore Data.” In Joint BeSHG / NVHG Meeting 2022, Abstracts.
Chicago author-date (all authors)
De Clercq, Griet, Bram Van Gaever, Lies Vantomme, Annelies Dheedene, and Björn Menten. 2022. “Benchmarking of Long-Read Structural Variant Callers Using Oxford Nanopore Data.” In Joint BeSHG / NVHG Meeting 2022, Abstracts.
Vancouver
1.
De Clercq G, Van Gaever B, Vantomme L, Dheedene A, Menten B. Benchmarking of long-read structural variant callers using Oxford Nanopore data. In: Joint BeSHG / NVHG meeting 2022, Abstracts. 2022.
IEEE
[1]
G. De Clercq, B. Van Gaever, L. Vantomme, A. Dheedene, and B. Menten, “Benchmarking of long-read structural variant callers using Oxford Nanopore data,” in Joint BeSHG / NVHG meeting 2022, Abstracts, Bruges, Belgium, 2022.
@inproceedings{8760521,
  abstract     = {{As long-read sequencing (LRS) technologies mature, several bioinformatics tools designed to identify structural variants (SVs) have been developed. To allow validation of these tools, Zook et al. published a highly curated SV truth set of Genome in a Bottle sample NA24385, consisting of deletions and insertions. We performed a benchmarking analysis with five LRS SV callers against this set utilising in-house generated Oxford Nanopore reads.
SVs are called with cuteSV, SVIM, sniffles, pbsv, and nanovar. The callers are assessed in terms of resource usage, reproducibility, and calling performance. The latter is evaluated with Truvari giving recall and precision statistics on SV detection. We further investigate the influence of read support, sequencing coverage, SV type and length, and integration of call sets with Jasmine.
CuteSV achieves overall best performance, while nanovar lags behind in both resource usage and calling statistics. A coverage greater than 20x offers no additional advantage for reliable SV detection, while the recommended read support of one third of the coverage proves to be too stringent. Integration of call sets with Jasmine should include three callers to compete with stand-alone call sets. 
We propose a minimum coverage of at least 15x for optimal sensitivity and specificity. Read support should be set at one fifth of the coverage to obtain optimised calling performance. CuteSV performs best in both sensitivity and specificity, and resource usage. Further work is however needed to assess results for different SV types and more complex regions.}},
  author       = {{De Clercq, Griet and Van Gaever, Bram and Vantomme, Lies and Dheedene, Annelies and Menten, Björn}},
  booktitle    = {{Joint BeSHG / NVHG meeting 2022, Abstracts}},
  language     = {{eng}},
  location     = {{Bruges, Belgium}},
  title        = {{Benchmarking of long-read structural variant callers using Oxford Nanopore data}},
  year         = {{2022}},
}