Run generate_pretrain_human.py in ./data/. Sequence length [1k, 5k, 10k] and numbers(100k) are required.
Run pretraining.py with the generated data. Configurations of different lengths shall be changed accordingly in config.yaml.
To fine-tune a pretrained model, you need to:
Run generate_data.py in ./data/ to save data. The sequence length is required. A total of 97,922 sequence will be extracted. ve_df.csv needs to be loaded.
Run classify_lightning.py to load the pretrained model under the folder ./human/100k_pretrain_best_epoch_9_8_16.pt and train cdilDNA.
When run classify_lightning.py to fine-tune a pretrained model, it also need config.yaml file in this script.
Could you please provide this file? or give some revision for classify_lightning.py.
Thank you!