remora
remora copied to clipboard
nanoporetech
Correct model to use for 5hmC calling
Not sure if this should be a bonito issue or a remora issue.
I run the following:
bonito basecaller [email protected] $input_path --modified-bases 5mC 5hmC --reference $reference > basecalls_with_mods.sam
I then I get the following error:
--- Logging error ---
Traceback (most recent call last):
File "/usr/local/lib/python3.8/dist-packages/remora/model_util.py", line 549, in load_model
submodels = submodels[modified_bases]
KeyError: '5hmc_5mc'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/lib/python3.8/logging/__init__.py", line 1085, in emit
msg = self.format(record)
File "/usr/lib/python3.8/logging/__init__.py", line 929, in format
return fmt.format(record)
File "/usr/local/lib/python3.8/dist-packages/bonito/mod_util.py", line 25, in format
self._style._fmt = self.fmt
AttributeError: 'CustomFormatter' object has no attribute 'fmt'
Call stack:
File "/usr/local/bin/bonito", line 8, in <module>
sys.exit(main())
File "/usr/local/lib/python3.8/dist-packages/bonito/__init__.py", line 34, in main
args.func(args)
File "/usr/local/lib/python3.8/dist-packages/bonito/cli/basecaller.py", line 75, in main
mods_model = load_mods_model(
File "/usr/local/lib/python3.8/dist-packages/bonito/mod_util.py", line 47, in load_mods_model
return load_model(
File "/usr/local/lib/python3.8/dist-packages/remora/model_util.py", line 551, in load_model
LOGGER.error(
Message: 'Remora model for modified bases 5hmc_5mc not found for [email protected].'
Arguments: ()
I checked the pre-trained models which suggest that at least for "v0.0.0" there should be a 5hmC model:
[15:29:15] Remora pretrained modified base models:
Pore Basecall_Model_Type Basecall_Model_Version Modified_Bases Remora_Model_Type Remora_Model_Version
---------------- --------------------- ------------------------ ---------------- ------------------- ----------------------
dna_r10.4_e8.1 fast 0.0.0 5mc CG 0
dna_r10.4_e8.1 fast 0.0.0 5mc CG 1
dna_r10.4_e8.1 fast 0.0.0 5hmc_5mc CG 0
dna_r10.4_e8.1 fast v3.3 5mc CG 0
dna_r10.4_e8.1 fast v3.3 5mc CG 1
dna_r10.4_e8.1 hac 0.0.0 5mc CG 0
dna_r10.4_e8.1 hac 0.0.0 5mc CG 1
dna_r10.4_e8.1 hac 0.0.0 5hmc_5mc CG 0
dna_r10.4_e8.1 hac v3.3 5mc CG 0
dna_r10.4_e8.1 hac v3.3 5mc CG 1
dna_r10.4_e8.1 sup 0.0.0 5mc CG 0
dna_r10.4_e8.1 sup 0.0.0 5mc CG 1
dna_r10.4_e8.1 sup 0.0.0 5hmc_5mc CG 0
dna_r10.4_e8.1 sup v3.4 5mc CG 0
dna_r10.4_e8.1 sup v3.4 5mc CG 1
dna_r10.4.1_e8.2 fast 0.0.0 5mc CG 0
dna_r10.4.1_e8.2 fast 0.0.0 5mc CG 1
dna_r10.4.1_e8.2 fast v3.5.1 5mc CG 0
dna_r10.4.1_e8.2 fast v3.5.1 5mc CG 1
dna_r10.4.1_e8.2 hac 0.0.0 5mc CG 0
dna_r10.4.1_e8.2 hac 0.0.0 5mc CG 1
dna_r10.4.1_e8.2 hac v3.5.1 5mc CG 0
dna_r10.4.1_e8.2 hac v3.5.1 5mc CG 1
dna_r10.4.1_e8.2 sup 0.0.0 5mc CG 0
dna_r10.4.1_e8.2 sup 0.0.0 5mc CG 1
dna_r10.4.1_e8.2 sup v3.5.1 5mc CG 0
dna_r10.4.1_e8.2 sup v3.5.1 5mc CG 1
I changed my call to:
bonito basecaller [email protected] $input_path --modified-bases 5mC 5hmC --reference $reference > basecalls_with_mods.sam
But this then complains about a non existent model. I am sure I am being stupid?
> available models:
- [email protected]
- [email protected]
- [email protected]
- [email protected]
- [email protected]
- [email protected]
- [email protected]
- [email protected]
- [email protected]
- [email protected]
- [email protected]
- [email protected]
The 0.0.0 model is meant to be the "default model" when selecting from the Remora repo, when the model version is not specified. But given that the specific Bonito version 3.4 is specified but does not have a trained model included Remora fails. This behavior should be adjusted to raise a warning and load the default model. I'll address this issue in the next Remora release. For now you can download the Remora repo and point Bonito to the 0.0.0 model (here in the Remora repo models/trained_models/dna_r10.4_e8.1/sup/0.0.0/5hmc_5mc/CG/v0/modbase_model.onnx) via the --modified-base-model Bonito argument.
This should be resolved with the latest 2.0 release. Please open this issue again if this issue persists.