Skip to content
DraconiticResearch
Search Sources API
197,768 studies31 sourcesUpdated 46 min ago
PreprintCC BY-NC-NDChemRxiv

Protein Language Model-Boosted Prediction of Isoelectric Point

  • Fangfang Luo1,2
  • Xiangxiang Lu1,2
  • Zhitao Cai1,2
  • Hongyu Lin1,2
  • Yandong Huang1,2
  1. 1College of Computer Engineering
  2. 2Jimei University
Open at sourcePDFDOICite

Abstract

pH plays a key role in many biological processes. When the solution pH equals the isoelectric point (pI) of a protein, the protein carries zero net charge, which minimizes its solubility and thus promotes aggregation. Here we present ESMpI, a sequence-based protein pI predictor that combines optimized pKa values of seven ionizable residue types and two terminals with embeddings from a pre-trained protein language model. Benchmarked against experimental protein pI values, ESMpI outperforms other state-of-the-art methods, including structure-based approaches that derive the pI of a protein from residue-specific pKa calculations. Finally, ESMpI requires only sequence information, making it well suited for high-throughput proteomescale analyses and protein engineering.

Related

  1. Understanding language model scaling for protein fitness prediction
    Chao Hou, Ди Лю, Aziz Zafar +1
    Nature Computational SciencePeer-reviewedLicence not statedFull text22 citedMolecular biology & gene regulationOpenAlex
  2. Protein–Protein Interface Residue Prediction by Integrating Protein Language Representations and Structural Graph Attention
    Zhen Hou, Wei He, Weinan Cao +4
    ACS OmegaPeer-reviewedCC BY-NC-ND 4.0Full textMolecular biology & gene regulationCrossrefEurope PMC
  3. AIPFusion-Net: an interaction-fusion framework for anti-inflammatory peptide prediction using protein language model embeddings, local sequence patterns, and handcrafted descriptors.
    Wentao Wang, Weidong Yang, Hongmei Li +2
    Frontiers in molecular biosciencesPeer-reviewedCC BYFull textMolecular biology & gene regulationEurope PMC
  4. ProteinJEPA: Latent prediction improves protein language model pretraining
    Dan Ofer, Dafna Shahaf, Michal Linial
    arXivPreprintCC BY-NC-SA 4.0Full textBiophysicsStatisticsarXiv
  5. SaltyMeta: a curated benchmark and protein language model-informed web tool for salty peptide prediction
    Wanchao Chen, Wen Li, Yanan He +1
    arXivPreprintCC BY-SA 4.0Full textBiophysicsarXiv

Cite

BibTeXRIS
BibTeX
@misc{luo2026protein,
  title = {{Protein Language Model-Boosted Prediction of Isoelectric Point}},
  author = {Fangfang Luo and Xiangxiang Lu and Zhitao Cai and Hongyu Lin and Yandong Huang},
  howpublished = {ChemRxiv},
  year = {2026},
  month = oct,
  doi = {10.26434/chemrxiv.15009857/v1},
  url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009857/v1},
  note = {CC BY-NC-ND},
}
RIS
TY  - UNPB
TI  - Protein Language Model-Boosted Prediction of Isoelectric Point
AU  - Fangfang Luo
AU  - Xiangxiang Lu
AU  - Zhitao Cai
AU  - Hongyu Lin
AU  - Yandong Huang
PY  - 2026
DA  - 2026/10/05
PB  - ChemRxiv
DO  - 10.26434/chemrxiv.15009857/v1
UR  - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009857/v1
AB  - pH plays a key role in many biological processes. When the solution pH equals the isoelectric point (pI) of a protein, the protein carries zero net charge, which minimizes its solubility and thus promotes aggregation. Here we present ESMpI, a sequence-based protein pI predictor that combines optimized pKa values of seven ionizable residue types and two terminals with embeddings from a pre-trained protein language model. Benchmarked against experimental protein pI values, ESMpI outperforms other state-of-the-art methods, including structure-based approaches that derive the pI of a protein from residue-specific pKa calculations. Finally, ESMpI requires only sequence information, making it well suited for high-throughput proteomescale analyses and protein engineering.
ER  -