An efficient consolidation of word embedding and deep learning techniques for classifying anticancer peptides:...

Area of Science:

• Biotechnology
• Computational Biology
• Bioinformatics

Background:

• Anticancer peptides (ACPs) offer a selective and safe alternative to conventional cancer therapies.
• The rapid increase in peptide sequences necessitates advanced computational models for accurate identification and classification.
• Peptide-based therapies aim to target cancer cells effectively while minimizing harm to normal cells.

Purpose of the Study:

• To develop and evaluate an efficient model for categorizing anticancer peptides (ACPs).
• To explore the efficacy of various word embedding techniques and deep learning architectures for ACP prediction.
• To establish a new state-of-the-art benchmark in ACP classification.

Main Methods:

• Evaluation of word embedding techniques: Word2Vec, GloVe, FastText, and One-Hot-Encoding for peptide sequence feature extraction.
• Application of deep learning models: Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Bidirectional LSTM (BiLSTM).
• Consolidation of embedding outputs with deep learning models for a hybrid classification framework.

Main Results:

• The FastText+BiLSTM model achieved 92.50% accuracy on the ACPs250 dataset.
• The same model demonstrated 96.15% accuracy on an independent dataset.
• The proposed framework significantly enhanced classification accuracy compared to existing state-of-the-art methods.

Conclusions:

• The FastText+BiLSTM model represents a highly effective approach for anticancer peptide classification.
• This study sets a new state-of-the-art performance in predicting anticancer peptides.
• The findings support the potential of integrating advanced computational methods for peptide-based cancer therapeutics development.