Protein Adsorption Predictions on Polymer Brushes with Chemical Embeddings

BB-EIT provides a generalized prediction model for protein adsorption amounts on polymer brush surfaces. The dataset, created by Shiwei Su and last updated in April 2026, integrates SMILES-based chemical representations with physicochemical features like polymer thickness and protein isoelectric point. Its 358.4 KB size indicates a focused collection of model-related data rather than a large experimental corpus.

Use Cases

• Predict protein adsorption amount using polymer SMILES strings and enumerated SMILES variants for data augmentation.
• Model adsorption behavior by integrating polymer features like water contact angle and surface charge with protein properties such as size and pI.
• Train or benchmark generalized interaction models on unseen polymer-protein systems using the provided physicochemical and biochemical feature set.
• Apply the BB-EIT framework to design antifouling materials by analyzing predictions for specific polymer brush and protein combinations.

Strengths

• Model leverages the pretrained ChemBERTa LLM architecture for robust chemical representation.
• Integrates a comprehensive set of features including polymer thickness, water contact angle, surface charge, protein pI, and size.

Limitations

• The 358.4 KB file size suggests a very limited scope, likely containing model parameters and feature sets rather than extensive experimental data.
• Lack of provided column names or sample data makes direct analysis of the underlying dataset structure impossible.
• Generalizability claims are model-focused; the underlying data's diversity and size for training are unknown.

Provenance

Source

figshare, authored by Shiwei Su.

Collection Method

Model development integrating pretrained ChemBERTa with augmented physicochemical and biochemical features.

Time Range

null

Freshness

Last updated in April 2026, indicating recent development.

Geography

null