Target-DBPPred: An intelligent model for prediction of DNA-binding proteins using discrete wavelet transform based...

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• DNA-protein interactions are fundamental to biological processes like transcription and recombination.
• DNA-binding proteins (DBPs) are implicated in diseases such as cancer and asthma, and are targets for drug development.

Purpose of the Study:

• To develop an intelligent computational method, Target-DBPPred, for accurate prediction of DNA-binding proteins (DBPs).
• To investigate novel feature descriptors for improved DBP prediction.

Main Methods:

• Employed a novel feature descriptor, EDF-PSSM-DWT (Evolutionary difference formula position-specific scoring matrix-discrete wavelet transform), alongside other multi-evolutionary methods (F-PSSM, EDF-PSSM, PSSM-DPC, Lead-BiPSSM).
• Utilized Sequential Forward Selection (SFS) for optimal feature set selection.
• Trained four classifiers: Adaboost, XGBoost, Extremely Randomized Trees (ERT), and Light XGBoost (LiXGB).

Main Results:

• The Light XGBoost (LiXGB) classifier, using the EDF-PSSM-DWT feature set, achieved the highest prediction accuracy.
• LiXGB demonstrated a performance improvement of 6.69% in training accuracy and 15.07% in testing accuracy compared to existing methods.
• The proposed Target-DBPPred predictor showed superior performance over current DBP prediction tools.

Conclusions:

• The developed Target-DBPPred method, particularly with the EDF-PSSM-DWT descriptor and LiXGB classifier, significantly enhances DBP prediction accuracy.
• This intelligent computational approach offers a valuable tool for advancing research in diseases associated with DBPs and for drug discovery.