A Comparative Study of QSPR Methods on a Unique Multitask PAMPA dataset

TL;DR

This study evaluates various QSPR models on a novel multitask PAMPA dataset, comparing their predictive performance and interpretability for membrane permeability prediction.

Contribution

It introduces a comprehensive multitask PAMPA dataset and systematically compares traditional descriptors with deep learning models for permeability prediction.

Findings

Physico-chemical descriptors outperform deep learning representations on small datasets.

The study highlights the trade-off between model interpretability and predictive accuracy.

It provides insights into membrane-specific permeability profiles across multiple organ models.

Abstract

We present a unique, multitask dataset comprising 143 drug and drug candidate molecules, each evaluated on in vitro, parallel artificial-membrane permeability assays (PAMPA) using six different model membranes. Using this resource, we systematically assess the effectiveness of various molecular descriptors and regression models in predicting passive membrane permeability. The studied models range from simple linear regression to a modern pre-trained transformer architecture. Particular attention is given to the trade-off between predictive performance and model interpretability, highlighting the challenges introduced by machine learning approaches. To our knowledge, this is the most comprehensive study on simultaneous modeling of multiple organ-specific PAMPA membranes to date, offering novel insights into membrane-specific permeability profiles. We found that expert-designed physico-chemical property descriptors are more fitting for a limited sample size permeabilty study than deep learning based representations.