Unifying pKa and Protonation Prediction with Sequence-based Deep Learning

pKaCHU: A Curated Microscopic pKa Dataset 

pKaCHU (data that are Combined, Honed, and Updated) is a curated dataset of experimentally derived microscopic pKa values with ionization state annotations for small molecules. 

Dataset Overview

9,000 microscopic pKa entries

Experimentally derived values

Explicit microstate pairs 

Canonicalized SMILES representations

Purpose

Experimental microscopic pKₐ data can be limited and inconsistently formatted across commonly used benchmarks. pKaCHU address these issues by:

Removing “acidic/basic” terminology 

Standardizing protonation and deprotonation conventions

Providing explicit microstate pair mappings

Harmonizing formatting across literature sources 

File Contents

This recode includes one dataset file (CSV format) with the following columns:

“prefix” – indicates whether the entry is associated with protonation (“Prot”) or deprotonation (“Deprot”)

“micropka input” – microstate pair in the format: protonated_SMILES>>deprotonated_SMILES

“pKas” – list of the reported pKₐs; some have more than one pKa due to overlapping from different sources

“average” – the average of the pKₐs if there is more than one pKₐ 

“citation” – the literature sources associated with the given molecule and pKₐ

Citation

If you use pKaCHU in your work, please cite:

Infante, C.; et al. Unifying pKₐ and Protonation Prediction with Sequence-Based Deep Learning. Manuscript under preparation (2026). DOI: 10.5281/zenodo.20089807

License 

This dataset is released under the Creative Commons Attribution 4.0 International license