PAMPA Membrane Permeability of 143 Drug Molecules Across Six Assays

A 2026 study by András Formanek presents a unique multitask dataset of 143 drug and drug candidate molecules. Each molecule was evaluated on in vitro parallel artificial membrane permeability assays (PAMPA) using six different model membranes. The resource was used to systematically assess the effectiveness of various molecular descriptors and regression models in predicting passive membrane permeability.

Use Cases

• Benchmarking regression models for permeability prediction based on the six PAMPA assay results.
• Comparing the performance of expert-designed physicochemical descriptors versus deep learning representations for a limited sample size.
• Studying the trade-off between predictive performance and model interpretability in membrane permeability modeling.
• Analyzing membrane-specific permeability profiles across different organ-specific model membranes.

Strengths

• Contains 143 unique drug and drug candidate molecules.
• Each molecule is evaluated across six different in vitro PAMPA model membranes, enabling multitask analysis.
• The study systematically compares a range of models from simple linear regression to a modern pretrained transformer architecture.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset size is 2.3 MB, indicating a relatively small-scale study.

Provenance

Source

figshare

Collection Method

In vitro parallel artificial membrane permeability assays (PAMPA) on drug molecules.

Freshness

Last updated 2026-04-25 04:14:34