EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

ThermoFormer: Predicting Protein Melting Temperature Through Large-Scale Pretraining

Jingchuan Li, Mingchen Li

Catalysts · 24.03.2026 · 16 (4) · 288

DOI: 10.3390/catal16040288 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Catalysts, online erschienen am 24.03.2026. Verfasst von Jingchuan Li und 1 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Computational Biology Key Program of Shanghai Science and Technology Commission, Science and Technology Innovation Key R&D Program of Chongqing.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Temperature plays a dominant environmental role in determining the efficiency of protein function. Accurately predicting protein thermal stability is crucial for fundamental biology, drug discovery, and protein engineering. Here, we introduce ThermoFormer, a transformer-based protein language model that learns both temperature-aware representation and sequence patterns. Specifically, we first built a large-scale dataset comprising more than 96 million protein sequences annotated with their optimal growth temperature (OGT). ThermoFormer is pre-trained with a supervised OGT prediction task and an unsupervised masked language modeling (MLM) task on the dataset. We evaluated ThermoFormer’s pre-training performance and its transferability to other temperature-prediction datasets, including two melting temperature (TM) datasets, an optimal catalytic temperature (OCT) dataset, and a thermophilic protein classification task. The results show that ThermoFormer achieves state-of-the-art performance across all evaluated tasks, outperforming prior unsupervised pre-trained models. In addition, we have also shown that ThermoFormer enables zero-shot temperature prediction, i.e., even without further fine-tuning, ThermoFormer can still achieve comparable performance. Our model can serve as a foundation for encoding protein sequences with temperature-aware representations, improving transferability to temperature-related downstream tasks.

Förderung

  • Computational Biology Key Program of Shanghai Science and Technology Commission
  • Science and Technology Innovation Key R&D Program of Chongqing

Zitieren

APA

Li, J., & Li, M. (2026). ThermoFormer: Predicting Protein Melting Temperature Through Large-Scale Pretraining. Catalysts, 16(4), 288. https://doi.org/10.3390/catal16040288

Vancouver

Li J, Li M. ThermoFormer: Predicting Protein Melting Temperature Through Large-Scale Pretraining. Catalysts. 2026;16(4):288. doi:10.3390/catal16040288

BibTeX
@article{Li2026,
  author = {Li, Jingchuan and Li, Mingchen},
  title = {ThermoFormer: Predicting Protein Melting Temperature Through Large-Scale Pretraining},
  journal = {Catalysts},
  year = {2026},
  volume = {16},
  number = {4},
  pages = {288},
  doi = {10.3390/catal16040288}
}
Kostenlos starten →