search

WGLab / DeepMod

DeepMod: a deep-learning tool for genomic-scale, strand-sensitive and single-nucleotide based detection of DNA modifications
Other
97 stars 35 forks source link

Compatibility with data got from using different versions of sequencing chemistry? #11

Closed ShangjinTan closed 5 years ago

ShangjinTan commented 5 years ago

The nanopore technology is evolving fast. I wondered if the trained model is compatible with data produced by using different versions of sequencing chemistry.

Musketeer-D commented 5 years ago

Yes, I also wonder will the trained model of deepmod be compatible with different sequencing kits and flow cells of ONT?

liuqianhn commented 5 years ago

Hi @ShangjinTan , @DAVID2022 , it is a good question. But I believe the effect of different sequencing chemistry is not very significant. There is one evidence in our paper: majority of e. coli data were sequenced by R9, while NA12878 were sequenced almost by R9.4. The cross-species testing show DeepMod achieved an accurate prediction.

But the version of basecallers may affect the performance, since I found that signals segmentations by different versions of basecallers contains some significant deviation. Thus, it would be cautious to work on datasets with different versions of basecallers.

Please feel free to share your idea and results about this question. Thank you.

Musketeer-D commented 5 years ago

Thank you very much for your kind help!