search

WGLab / NanoCaller

Variant calling tool for long-read sequencing data
MIT License
90 stars 8 forks source link

Models compatible with Guppy6.x.x #31

Open sagnikbanerjee15 opened 1 year ago

sagnikbanerjee15 commented 1 year ago

Hello,

Are the error models compatible with the latest version of guppy? If not, which models would you suggest for r9.4.1 and r.10.1?

Thanks

umahsn commented 1 year ago

Hi,

Generally speaking, the models are compatible across Guppy versions with minor performance differences. For R9.4.1, the model ONT-HG002_bonito is probably best suited for Guppy6.x.x reads since Bonito basecalling models were incorporated into Guppy v5 and onward. This model was trained using R9.4.1 reads basecalled with Bonito 0.30. As for R10.1 reads, I would recommend ONT-HG002_r10.3. However, for either datasets, you can also use the default model ONT-HG002 which we have shown generalizes well to different flowcell types and basecallers.

The recommendations above are for SNP models. For indel models, the choice of models is not very critical as long as you use one of the two ONT models ONT-HG001 and ONT-HG002.