/LMSuccSite

Improving Protein Succinylation Sites Prediction Using Features Extracted from Protein Language Model

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

LMSuccSite

Improving Protein Succinylation Sites Prediction Using Features Extracted from Protein Language Model

Webserver

http://kcdukkalab.org/LMSuccSite/

Authors

Suresh Pokharel1, Pawel Pratyush1, Michael Heinzinger2, Robert H. Newman3, Dukka B KC1*
1Department of Computer Science, Michigan Technological University, Houghton, MI, USA
2Department of Informatics, Bioinformatics and Computational Biology - i12, TUM (Technical University of Munich), Boltzmannstr. 3, 85748, Garching/Munich, Germany
3Department of Biology, College of Science and Technology, North Carolina A&T State University, Greensboro, NC, USA

* Corresponding Author: dbkc@mtu.edu

Installation

Clone the repository: git clone git@github.com:KCLabMTU/LMSuccSite.git or download https://github.com/KCLabMTU/LMSuccSite

Install Libraries

Python version: 3.9.7

Install from requirement.txt: pip install -r requirements.txt

Required libraries and versions: Bio==1.5.2 keras==2.9.0 matplotlib==3.5.1 numpy==1.23.5 pandas==1.5.0 requests==2.27.1 scikit_learn==1.2.0 seaborn==0.11.2 tensorflow==2.9.1 torch==1.11.0 tqdm==4.63.0 transformers==4.18.0 xgboost==1.5.0

Install Transformers

pip install -q SentencePiece transformers

Model evaluation using the existing benchmark independent test set

Please run the evaluate_model.py script. To evaluate our model on the independent test set, we have already placed the test sequences and corresponding ProtT5 features in data/test/ folder. Once you install the requirements, run the following command:
python evaluate_model.py

To run LMSuccSite model on your own sequences

In order to predict succinylation site using your own sequence, you need to have two inputs:

  1. Copy sequences you want to predict to input/sequence.fasta
  2. Run python predict.py
  3. Find results inside output folder

Training and other experiments

  1. Find training data at data/train/ folder
  2. Find all the codes and models related to training at training codes folder.

Citation

Pokharel, S., Pratyush, P., Heinzinger, M. et al. Improving protein succinylation sites prediction using embeddings from protein language model. Sci Rep 12, 16933 (2022). https://doi.org/10.1038/s41598-022-21366-2

Link: https://rdcu.be/cXFfM

Contact

Please send an email to sureshp@mtu.edu (CC: dbkc@mtu.edu) for any kind of queries and discussions.

Webserver

http://kcdukkalab.org/LMSuccSite/