SkipCPP-Pred: an improved and promising sequence-based predictor for predicting cell-penetrating peptides.

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology and Cheminformatics

Background:

• Cell-penetrating peptides (CPPs) are short amino acid sequences (5-30 residues) capable of translocating across cell membranes with high efficiency.
• CPPs hold significant therapeutic potential for applications in gene therapy and cancer treatment due to their delivery capabilities.
• Existing computational methods for CPP prediction lack the necessary accuracy and reliability for widespread application.

Purpose of the Study:

• To develop a novel and improved computational predictor for cell-penetrating peptides (CPPs).
• To enhance the accuracy and reliability of CPP identification for accelerating therapeutic applications.
• To introduce a new sequence-based feature representation for capturing residue correlations.

Main Methods:

• Development of SkipCPP-Pred, a sequence-based predictor utilizing an adaptive k-skip-n-gram algorithm for feature representation.
• Integration of adaptive skip features with a Random Forest (RF) classifier to build the prediction model.
• Creation of a high-quality benchmark dataset by minimizing redundancy and improving class similarity to mitigate performance bias.

Main Results:

• SkipCPP-Pred achieved an accuracy 3.6% higher than current state-of-the-art CPP predictors, as validated by jackknife tests.
• The adaptive k-skip-n-gram model effectively captures intrinsic residue correlation information.
• The refined benchmark dataset yielded a promising predictive model, avoiding biases present in existing methods.

Conclusions:

• SkipCPP-Pred is a simple, fast, and accurate sequence-based predictor for cell-penetrating peptides.
• The adaptive k-skip-n-gram model is key to the enhanced predictive performance.
• SkipCPP-Pred is accessible via a public webserver for broader research use.