The Challenges of PROTAC Permeability Prediction

2026-10-06

AI Summary

This study addresses the challenge of predicting PROTAC cell permeability, which is hindered by the scarcity and inconsistency of public data. We propose an expert-in-the-loop literature mining pipeline leveraging large language models, integrated with optical chemical structure recognition (OCSR) and manual verification, to expand a public PAMPA dataset from 31 to 87 entries. A predictive model is subsequently constructed using Ridge regression. Our findings reveal that data composition, rather than sheer volume, is the critical factor constraining model generalizability. Notably, the model trained on the augmented data demonstrates excellent backward transfer performance (ρ=0.80). This work provides an effective paradigm for overcoming the bottleneck of PROTAC permeability prediction under small-sample conditions.

Abstract

Cell permeability is a key bottleneck for PROTAC development, and public data available to model it is scarce and inconsistent. We adapt an expert-in-the-loop LLM extraction workflow to mine PAMPA measurements from the primary literature, recovering image-only structures by optical chemical structure recognition and hand-verifying every record, expanding the public record from 31 PROTACs to 87. Ridge models trained on PROTAC-DB 3.0 reach $R^2 = 0.67$ within that resource but collapse on the newly extracted chemistry ($ρ= 0.12$), while models trained on the new compounds transfer back successfully ($ρ= 0.80$). We conclude that the current composition of the published records, and not dataset size, is limiting the construction of more generalizable models, and we outline what would need to change in reporting practices for better data-driven permeability models.

Problem

• PROTAC
• cell permeability prediction
• data scarcity
• model generalizability
• PAMPA

Innovation

• PROTAC permeability prediction
• expert-in-the-loop LLM extraction
• optical chemical structure recognition
• model generalizability
• PAMPA data mining