Hi everyone,
I am an undergraduate student trying to understand the working of Apta-MCTS (https://pmc.ncbi.nlm.nih.gov/articles/PMC8232527/). I believe that initially, I have to run the preprocess.py file first and then classifier.py for RNA aptamer classification.
Problem 1: I assumed that preprocess.py would generate files called train.json and test.json, which are required to run classifier.py, but preprocess.py does not seem to generate any output files.
Problem 2: I tried to convert the data from excel files referenced by the authors into .json files using the template provided in their GitHub (https://github.com/leekh7411/Apta-MCTS). (Just to check the working of classifier.py)
I have two Excel files containing information about proteins and aptamers and I need to structure the JSON output as follows:
{
"targets": {
"<protein_name>":{
"model": {
"method" : "Lee_and_Han_2019|Apta-MCTS",
"score_function" : "<path of the weights of the pre-trained API classifer>",
"k" : "<number of top scored candidates>",
"bp" : "<length of candidate RNA-aptamer sequences>",
"n_iter" : "<number of iterations for each base when method is Apta-MCTS>"
},
"protein": {
"seq" : "<target protein sequence>"
},
"aptamer": {
"name" : [],
"seq" : []
},
"candidate-aptamer": {
"score" : [],
"seq" : [],
"ss" : [],
"mfe" : []
},
"protein-specificity": {
"name" : "<list of name of proteins that do not want to bind>",
"seq" : "<list of sequence of proteins that do not want to bind>"
}
}
},
"n_jobs" : "<number of available cores for the multiprocessing tasks>"
}
However, the resulting JSON does not match the expected format, causing classifier.py to throw a KeyError: 'protein-seq'.
Input: python3 classifier.py -dataset_dir=datasets/li2014 -tag=rf-iCTF-li2014 -min_trees=35 -max_trees=200 -n_jobs=20 -num_models=1000
Error:
dataset_dir=datasets/li2014 -tag=rf-iCTF-li2014 -min_trees=35 -max_trees=200 -n_jobs=20 -num_models=1000
Traceback (most recent call last):
File "/home/cake13/Apta-MCTS/paper_version/classifier.py", line 131, in <module>
fire.Fire(main)
File "/home/cake13/ViennaRNA-2.7.0/vienna_env/lib/python3.12/site-packages/fire/core.py", line 135, in Fire
component_trace = _Fire(component, args, parsed_flag_args, context, name)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/cake13/ViennaRNA-2.7.0/vienna_env/lib/python3.12/site-packages/fire/core.py", line 468, in _Fire
component, remaining_args = _CallAndUpdateTrace(
^^^^^^^^^^^^^^^^^^^^
File "/home/cake13/ViennaRNA-2.7.0/vienna_env/lib/python3.12/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace
component = fn(*varargs, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^
File "/home/cake13/Apta-MCTS/paper_version/classifier.py", line 119, in main
trainset = load_benchmark_dataset(train_json_path)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/cake13/Apta-MCTS/paper_version/preprocess.py", line 243, in load_benchmark_dataset
pseqs = d["protein-seq"]
~^^^^^^^^^^^^^^^
KeyError: 'protein-seq'
Questions:
- Could there be an issue with how I structured the JSON from Excel?
- Are there any best practices for formatting Excel-to-JSON conversions? Is that something that can be done or is my understanding of a json file wrong?
- Any suggestions for debugging where the JSON format might be incorrect?
- Do I need any additional files that need to be created or sourced from somewhere apart from what is provided by the authors in their GitHub (https://github.com/leekh7411/Apta-MCTS)?
Thanks in advance for any help!
