Bonito research model, dramatic increase in basecall quality scores

[callumparr]

I ran bonito on some old 9.4.1 DNA libraries and the quality scores seems dramatically better and tighter than flip flop model. Is this expected?

[iiSeymour]

@callumparr looks reasonable - what version of guppy are the flip flop results from and do you have alignment accuracies to compare against?

[callumparr]

This was actually an older version from 2019 but I rerun with 4.4.1 and the qualities seem similar. Just looking at accuracies now.

I did see that bonito consistently gives shorter lengths than flip flop. Is this also expected? Ultimately I want to use better basecalling to help with demultiplexing single-cell RNA seq data sets so ideally want to make sure have the ends are of good quality

[crysclitheroe]

Hi, we are installing Guppy, Rerio and bonito basecallers on our gpu cluster, but are wondering how often latest bonito models get updated to Rerio, and where/how we can check this?

Thanks and kind regards Crystal

[davidnewman02]

Closed as this model has now been superseded.