Local Dorado models are not being recognised
Closed this issue · 1 comments
Hello,
We have a somewhat restrictive IT security environment around our HPC and I can download and save the models using a machine from outside the network that hosts the HPC we have access to, and I can put the saved models in whatever directory is needed by uploading the models I saved previously elsewhere. Despite supplying the models in /bin /dorado or /dorado-models, I keep getting the following error when calling Dorado like so:
./dorado basecaller --emit-fastq dna_r10.4.1_e8.2_400bps_fast@v4.1.0 /scratch/<user_specific_path>/PBL_test_pod5 > dorado_log_20240516.log
[2024-05-16 09:11:46.875] [info] Running: "basecaller" "--emit-fastq" "dna_r10.4.1_e8.2_400bps_fast@v4.1.0" "/scratch/<user_specific_path>/PBL_test_pod5"
[2024-05-16 09:11:46.894] [info] - Note: FASTQ output is not recommended as not all data can be preserved.
[2024-05-16 09:11:46.894] [info] > Creating basecall pipeline
[2024-05-16 09:11:46.894] [error] toml::parse: file open error -> dna_r10.4.1_e8.2_400bps_fast@v4.1.0/config.toml
I understand the desire of ONT to have tight control over models to ensure ONT products are performing as intended, but is there a way to get around this and force Dorado to find the models in a specific location? Attempting to download the models never works (and wont) as getting a cert to do so on our HPC is not easy/common/permitted.
Best wishes
Eddie
Hi,
You can use models in any location by passing the path of the model folder to dorado instead of the model name.