Quick start guide¶
Once you have succesfully installed the package, you can start using it.
Downloading the dataset¶
For this tutorial we will work with the GLASS dataset
To download it, we will use the following code:
python workflow/scripts/get_datasets.py glass --min_num_aa 250
Since this is a dataset of GPCRs, we do not accept protein structures with less than 250 aminoacids.
This will create a new folder in datasets/glass/resources, which will contain all the necessary information.
Running the Snakemake pipeline¶
Once our dataset is downloaded, we can run the snakemake pipline with the following command:
snakemake -j 10 --use-conda --configfile config/snakemake/glass.yaml
This will create the final pickle file for the GLASS dataset, which will be located in datasets/glass/resources/prepare_all folder.
Note
Some parts of the config rely on more than just the base file downloaded in the previous step.
Running the training¶
We can run the training script with the following code:
python train.py config/dti/glass.yaml
This will start the training process, which will take a while.
You can monitor the progress of the training by running the following command:
tensorboard --logdir=tb_logs