SigmaFormer : Augmenting transformer encoders with COSMO sigma profiles for pure component property prediction
- 1William G. Lowrie Department of Chemical and Biomolecular Engineering The Ohio State University Columbus Ohio USA
- 2Artie McFerrin Department of Chemical Engineering Texas A&M University College Station Texas USA
Abstract
Transformer‐based molecular models pretrained on SMILES strings demonstrate strong performance in property prediction. However, these model often lack explicit integration of molecular surface charge distributions that govern intermolecular interactions such as hydrogen bonding and polarity. SigmaFormer addresses this limitation by augmenting a pretrained encoder with a 53‐dimensional descriptor derived from COSMO quantum‐chemical calculations, including the σ‐profile, cavity surface area, and cavity volume. Benchmarking across 25 thermophysical and environmental properties indicates that SigmaFormer achieves the highest average (0.836) and the best in 9 of 25 properties among five models. Performance improvements are most pronounced in properties governed by intermolecular interactions, with phase transition and environmental/toxicity categories exhibiting average of +2.30% and +1.03%, respectively. Monte Carlo dropout uncertainty is reduced in 19 of 25 properties. Data‐efficiency experiments identify three distinct contribution modes, and gradient‐based interpretability analysis demonstrates that the model leverages σ‐profile regions in a thermodynamically consistent and property‐specific manner.
Related
- Geometry-Aware Line Graph Transformer Pretraining for Molecular Property Prediction
- Transformer-based molecular fragment prediction using SMILES and DeepSMILES representations in a fragment-based drug discovery pipeline.
- ANN-based prediction of torsional fatigue behavior in an injection-molded compliant mechanism component
- Frozen sequence encoders with gradient boosting match the state of the art on drug–target affinity domain generalization, and a sequence defect in the Davis kinase benchmark
- Augmenting Middle‐Down Sequence Coverage With Trapped Ion Mobility Spectrometry
BibTeX
@article{kim2026sigmaformer,
title = {{SigmaFormer : Augmenting transformer encoders with COSMO sigma profiles for pure component property prediction}},
author = {Tae Hyun Kim and Silabrata Pahari and Joseph Sang‐Il Kwon},
journal = {AIChE Journal},
year = {2026},
month = may,
doi = {10.1002/aic.70450},
url = {https://doi.org/10.1002/aic.70450},
note = {CC BY},
}
RIS
TY - JOUR
TI - SigmaFormer : Augmenting transformer encoders with COSMO sigma profiles for pure component property prediction
AU - Tae Hyun Kim
AU - Silabrata Pahari
AU - Joseph Sang‐Il Kwon
PY - 2026
DA - 2026/05/17
JO - AIChE Journal
DO - 10.1002/aic.70450
UR - https://doi.org/10.1002/aic.70450
AB - Transformer‐based molecular models pretrained on SMILES strings demonstrate strong performance in property prediction. However, these model often lack explicit integration of molecular surface charge distributions that govern intermolecular interactions such as hydrogen bonding and polarity. SigmaFormer addresses this limitation by augmenting a pretrained encoder with a 53‐dimensional descriptor derived from COSMO quantum‐chemical calculations, including the σ‐profile, cavity surface area, and cavity volume. Benchmarking across 25 thermophysical and environmental properties indicates that SigmaFormer achieves the highest average (0.836) and the best in 9 of 25 properties among five models. Performance improvements are most pronounced in properties governed by intermolecular interactions, with phase transition and environmental/toxicity categories exhibiting average of +2.30% and +1.03%, respectively. Monte Carlo dropout uncertainty is reduced in 19 of 25 properties. Data‐efficiency experiments identify three distinct contribution modes, and gradient‐based interpretability analysis demonstrates that the model leverages σ‐profile regions in a thermodynamically consistent and property‐specific manner.
SN - 0001-1541
SN - 1547-5905
ER -