kkyamada / bert-rbp Goto Github PK

I downloaded DNABERT3, finetuned the model for TIAL1 (Step 3.2) and everything worked fine. But when I try to start the attention analysis with source analyze_regiontype.sh TIAL1 it gives an error stating:

05/08/2022 13:50:01 - INFO - __main__ - Training/evaluation parameters Namespace(cache_dir='', config_name='', data_dir='../sample_dataset/TIAL1/nontraining_sample_finetune/hg38/', device=device(type='cuda'), do_analyze_regiontype=True, do_analyze_regiontype_specific=False, do_cache=False, do_lower_case=False, eval_all_checkpoints=False, fp16=False, fp16_opt_level='O1', local_rank=-1, logging_steps=500, max_seq_length=101, model_name_or_path='../sample_dataset/TIAL1/finetuned_model', model_type='dna', n_gpu=1, n_process=8, no_cuda=False, output_mode='classification', output_visimage=False, overwrite_cache=False, per_gpu_pred_batch_size=64, predict_dir='../sample_dataset/TIAL1/finetuned_model/analyze_regiontype/', rbp_name='', region_type=0, save_steps=500, save_total_limit=None, seed=42, server_ip='', server_port='', should_continue=False, specific_heads='', task_name='dnaprom', tokenizer_name='', visualize_models=None) Traceback (most recent call last): File "../examples/run_analysis_regiontype.py", line 876, in <module> main() File "../examples/run_analysis_regiontype.py", line 798, in main output_dir = args.output_dir.replace("/690", "/690/" + str(kmer)) AttributeError: 'Namespace' object has no attribute 'output_dir'

But when I check the run_analysis_regiontype.py script for the output_dir argument, i realised it is never set in this script. How can i set the output dir without manipulating your code? Am I missing something?

Thanks in advance :)

Hi.Is the second point in the sixth section of this annotations.ipynb file missing something? I can't convert the file to .csv format.

Hello, I'm trying to install and run your project, but I'm getting the following error when I run your installation instructions (in step 1.1):

$ python -m pip install --editable .
Obtaining file:///C:/Users/megan/Code/bert-rbp
ERROR: file:///C:/Users/megan/Code/bert-rbp does not appear to be a Python project: neither 'setup.py' nor 'pyproject.toml' found.

Is there a missing file, or am I doing something wrong? Thanks!

What should I do if I want to implement the step of Data preprocessing on the window system? I have downloaded the data set and I have tried set benchmark _ path = c:/users/LGD/Bert-RBP-master/benchmark.gz.
set OUTPUT_PATH=./datasets
python generate_datasets.py --path_to_benchmark %BENCHMARK_PATH% --path_to_output %OUTPUT_PATH%
This command, but my folder is still empty.

Hi there,

Thank you for attention.
I am doing a work, applying your BERT-RBP and the DNABERT models on some plant gene inputs, while the program dropped errors about keyerror of the plant gene inputs.
May I ask if I could directly use plant gene and sequence as inputs to your model?

Wish you a lovely day.

Best regards,
Chao

Hi, are the fine-tuned models for the ENCODE RBPs available for download somewhere?

Hi @kkyamada ,

Thanks for sharing this great work!

I am curious how you computed AUROC scores of those benchmarking Binding Sites prediction Models. I assume your AUROC score is for bound or not (whole sequence), yet those models predict bound or not for each site.

Thanks!