DNA-Prot: identification of DNA binding proteins from protein sequence information using random forest

Area of Science:

• Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• DNA-binding proteins (DNABPs) are crucial for fundamental cellular processes including DNA replication, repair, and transcriptional regulation.
• Current methods for identifying DNABPs primarily rely on protein structure, with limited options available for sequence-based identification.

Purpose of the Study:

• To develop and evaluate a novel random forest-based method, termed DNA-Prot, for the accurate identification of DNA-binding proteins directly from their amino acid sequences.
• To compare the performance of DNA-Prot against existing methods like DNAbinder using various benchmark datasets.

Main Methods:

• A random forest classification algorithm was trained and tested using curated datasets of known DNA-binding and non DNA-binding proteins.
• Performance evaluation involved calculating accuracy on training, testing, and independent benchmark datasets.
• Comparative analysis was conducted against the DNAbinder method using sequence-derived features.

Main Results:

• DNA-Prot achieved high accuracy, reaching 80.31% on the training set and 84.37% on the test set.
• Benchmarking on a large independent dataset (823 DNABPs, 823 non-DNABPs) demonstrated DNA-Prot's superior performance (81.83% accuracy) compared to DNAbinder (61.42%-63.5%).
• DNA-Prot consistently outperformed DNAbinder on additional benchmark datasets, highlighting its effectiveness.

Conclusions:

• DNA-Prot is an efficient and accurate computational tool for identifying DNA-binding proteins based solely on protein sequence information.
• The developed method provides a valuable resource for researchers studying protein function and cellular mechanisms.
• The DNA-Prot software and datasets are publicly available for further research and application.