search

nanoporetech / dorado

Oxford Nanopore's Basecaller
https://nanoporetech.com/
Other
439 stars 53 forks source link

run multi model dorado with no internet connection #914

Open gambalab opened 3 days ago

gambalab commented 3 days ago

I am using Dorado on a cluster where individual nodes do not have internet access, only the front-end node does. Following the tutorial, I have downloaded all the models and can run Dorado by specifying the offline model to use. However, when I run Dorado on my local machine and use automatic model selection for duplex calling, Dorado downloads and uses three models, as shown below.

How can I achieve the same result using the offline models I have downloaded?

[2024-06-23 13:14:55.085] [info] Running: "duplex" "sup" "--device" "cuda:0,1" "--emit-fastq" "mnt/HD10TB/tmp/HG002_Duplex/POD5/1_3_23_R1041_Duplex_HG002_2/"
[2024-06-23 13:14:55.088] [info] > No duplex pairs file provided, pairing will be performed automatically
[2024-06-23 13:14:55.088] [info]  - Note: FASTQ output is not recommended as not all data can be preserved.
[2024-06-23 13:14:56.684] [info]  - downloading dna_r10.4.1_e8.2_400bps_sup@v4.1.0 with httplib
[2024-06-23 13:14:59.212] [info]  - downloading dna_r10.4.1_e8.2_4khz_stereo@v1.1 with httplib
[2024-06-23 13:14:59.413] [info]  - downloading dna_r10.4.1_e8.2_4khz_stereo@v1.1 with httplib
HalfPhoton commented 2 hours ago

Please place the dorado models in the current working directory.

[!NOTE] You can achieve this with symbolic links (ln -s) to avoid copying the models.

This requirement will change in an upcoming release as I've added a --models-directory <PATH> argument to dorado to add a search directory for automatic model selection.

Kind regards, Rich