pSuc-FFSEA

pSuc-FFSEA is a predictive tool designed to identify lysine succinylation sites in protein sequences. Succinylation is a protein post-translational modification that plays a crucial role in regulating protein conformation and cellular functions. This software tool aims to facilitate the identification of succinylation sites, which are associated with the development of various diseases.

Key features of pSuc-FFSEA:
1. Feature Fusion: pSuc-FFSEA extracts sequence information and physicochemical properties from protein sequences using various methods, such as EBGW, One-Hot encoding, continuous bag-of-words, chaos game representation, and AAF_DWT.
2. Feature Selection: LASSO (Least Absolute Shrinkage and Selection Operator) is applied to select the most informative features for the classifier.
3. Stacking Ensemble Algorithm: pSuc-FFSEA employs a two-layer stacking ensemble approach, using three base classifiers (SVM, broad learning system, and LightGBM) in the first layer, and a logistic regression classifier as the meta-classifier in the second layer.
4. Model Optimization: The tool optimizes classifier hyperparameters using the Bayesian optimization algorithm and grid search algorithm to improve prediction accuracy and reduce computational efforts.
5. Robust Performance: pSuc-FFSEA demonstrates robust performance with an average prediction accuracy of 0.7773 ± 0.0120 based on rigorous 10-fold cross-validation.