A hybrid approach for predicting transcription factors

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• Transcription factors are crucial DNA-binding proteins regulating gene expression and cellular processes.
• Accurate prediction of transcription factors is vital for understanding cell differentiation, signaling, and cell-cycle control.

Purpose of the Study:

• To develop a highly accurate hybrid method for predicting transcription factors.
• To combine the strengths of alignment-based and alignment-free approaches for improved prediction performance.

Main Methods:

• Developed alignment-free models using machine learning and protein composition features.
• Employed BLAST for alignment-based transcription factor prediction.
• Created a hybrid model by integrating scores from both alignment-free and alignment-based methods.

Main Results:

• The best alignment-free model achieved an AUC of 0.97 on an independent dataset.
• The hybrid method, combining alignment-free and alignment-based scores, reached a maximum AUC of 0.99.
• The proposed hybrid method demonstrates superior performance compared to existing prediction tools.

Conclusions:

• The hybrid TransFacPred method offers a significant advancement in transcription factor prediction accuracy.
• The developed models are available via a webserver, Python Package Index, and standalone package for broader accessibility.