Nanopore pipeline fails when specifying medaka model r103_hac_g507

[thekatpod]

Hello, thanks for a great contribution to the ONT community. I'm having issues with the pipeline failing because it seems like it's using an older version of medaka which does not include the new models for Guppy version 5.0.7. I've basecalled my data using Guppy 5, so I'd like to use a corresponding medaka model. Could this be updated or am I doing something wrong? I am unsure what the consequences of polishing with medaka using a model based on 3.4.5 when I've basecalled with 5.0.7. Thanks in advance!

[SorenKarst]

Hi,

Thank you.

It is using an older version of Medaka, but it might work if you try to update it manually.

Install the development version of the pipeline: https://github.com/SorenKarst/longread_umi/tree/develop

Run the following code

conda update -n longread_umi medaka

Let me know how it goes.

regards
Søren

[thekatpod]

Hi Soren. Thanks so much for you response. I downloaded the developer version, and it installed fine, ran tests etc. and all good. I ran the update code, but the version of Medaka is still 1.0.1 unfortunately.

[SorenKarst]

The newer versions of medaka has some dependencies that clashes with the some of the older packages used in the pipeline.

I will look at a fix.

[cfblaeb]

Upgrading the dependencies to work with newer models would very much be appreciated

[MaestSi]

Hi @thekatpod and @cfblaeb ,
please have a look at this post for testing a longread_umi-inspired pipeline with updated dependencies.
SM