POSEIDON

POSEIDON predicts quantitative cellular uptake of cell-penetrating peptides (CPPs) across cell lines using machine-learning regression to support design and optimization of CPP-based delivery systems.

Key Features:

• Curated Database: Provides an open-access curated database with experimental quantitative uptake values for over 2,300 entries and physicochemical properties for 1,315 peptides, including cell line specificity, cargo types, and peptide sequences.
• Machine Learning Regression Predictor: Integrates the POSEIDON database with genomic features of various cell lines to train regression models that predict CPP uptake across cell lines, reporting Pearson r = 0.87, Spearman ρ = 0.88, and r² = 0.76 on independent test sets.

Scientific Applications:

• CPP design and optimization: Enables design and optimization of CPPs with enhanced specificity and uptake efficiency using quantitative uptake data and regression-based predictions.
• Prioritization for experimental testing: Reduces the need for labor-intensive in vitro/in vivo validation by prioritizing candidate CPPs based on predicted uptake across cell lines.
• Therapeutic peptide delivery development: Supports development and improvement of therapeutic peptide delivery systems by combining comprehensive datasets with predictive modeling.

Methodology:

Uses machine-learning regression models trained on a curated database of quantitative uptake values and peptide physicochemical properties integrated with genomic features of cell lines.