This repository is to predict 3 state 2nd structure of a protein.
cd ./2nd_stru/ #Bash
- to use the predictor successfully, you must in /2nd_stru/ folder
- If you want to run model built by sequence information.
python3 ./script/input_seq.py ./input/input.txt #Bash
- Attention:
- You can change your own test file path in <input sequence file>
- If you want to run model built by pssm.
python3 ./script/input_pssm.py ./input/pssm/ #Bash
- Attention:
- You can change your own pssm path in <input pssm folder>, path folder must end with '/'
- The pssm filename must starts with '>' and end with '.pssm'. For example: >1hnf-1-AS.pssm
- ./input/
- ./logs/model_pssm/
- ./logs/model_seq/
- ./script/
- ./output/
- Output is in /2nd_stru/output/
- ./output.txt
- Prediction by sequence model.
- Format: 3line.
- First line: >ID
- Second line: amino acid sequences
- Third line: Topology sequences
- ./output_pssm.txt
- Prediction by pssm model.
- Format: 2line
- First line: >ID
- Second line: Topology sequences
- You can conbine your sequences and predicted topology on your own.
contain a sequence to test
the script to run
functions to extract data structures, sliding window, and onehotencode.
manually split dataset into 5-fold.
manually do cross-validation and change parameter
models and arrays produced during building the predictor.
contain the model I have built.
SVM input array. Use in cross-validation
contain the output file.