Joint protein-DNA language modeling enables prediction and engineering of DNA recognition specificity
- 1Peking University
Abstract
Protein-DNA recognition specificity underlies transcriptional regulation, DNA replication, chromatin remodeling and genome editing. Protein-DNA recognition depends on both interacting sequences, yet language models that learn this cross-molecular dependence explicitly during pretraining remain limited. Here, we developed the Protein-DNA Language Models (PDLMs), interaction-aware language models that jointly represent protein and DNA sequences. PDLMs were built by continued pretraining from ESM2 with masked language modeling on paired protein-DNA sequences. A shared encoder places amino acids and nucleotides in the same molecular context, enabling conditional sequence scoring and transferable pair representations. We further introduced a domain-aware masking strategy that increased masking probability within DNA-binding domains to strengthen learning of interaction-relevant regions. Across downstream tasks, PDLMs improved protein-DNA pair classification relative to independent-encoder baselines and supported prediction of mutation-associated binding changes. Zero-shot scoring recovered known changes in DNA-recognition specificity and prioritized variants with higher editing activity. Learned representations also supported cross-study prediction of base editor activity. After fine-tuning on transcription factor binding data, PDLMs predicted DNA motifs directly from protein sequences. Together, PDLMs provide a framework for interaction-aware biomolecular language modeling, with potential applications in protein engineering and genome editing.
Related
- Limitations of general protein language models and public TCRpMHC data in training specificity prediction models
- Antibody-Antigen Affinity Prediction with Chain-Aware Protein Language Modeling
- TraSPIN: inferring transporter specificity through language model-guided graph learning
- Evidence-resolved benchmarks separate biological context recognition from direct-partner specificity in protein interaction models
- Engineering the effector specificity of the macrolide-sensing transcription factor MphR for the detection of sesquiterpenes
BibTeX
@misc{xiong2026joint,
title = {{Joint protein-DNA language modeling enables prediction and engineering of DNA recognition specificity}},
author = {Y. Xiong and J. Ren and X. Wei and X. Li and J. Li and W. Wei},
howpublished = {bioRxiv},
year = {2026},
month = nov,
doi = {10.64898/2026.11.05.763984},
url = {https://www.biorxiv.org/content/10.64898/2026.11.05.763984v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Joint protein-DNA language modeling enables prediction and engineering of DNA recognition specificity
AU - Y. Xiong
AU - J. Ren
AU - X. Wei
AU - X. Li
AU - J. Li
AU - W. Wei
PY - 2026
DA - 2026/11/05
PB - bioRxiv
DO - 10.64898/2026.11.05.763984
UR - https://www.biorxiv.org/content/10.64898/2026.11.05.763984v1
AB - Protein-DNA recognition specificity underlies transcriptional regulation, DNA replication, chromatin remodeling and genome editing. Protein-DNA recognition depends on both interacting sequences, yet language models that learn this cross-molecular dependence explicitly during pretraining remain limited. Here, we developed the Protein-DNA Language Models (PDLMs), interaction-aware language models that jointly represent protein and DNA sequences. PDLMs were built by continued pretraining from ESM2 with masked language modeling on paired protein-DNA sequences. A shared encoder places amino acids and nucleotides in the same molecular context, enabling conditional sequence scoring and transferable pair representations. We further introduced a domain-aware masking strategy that increased masking probability within DNA-binding domains to strengthen learning of interaction-relevant regions. Across downstream tasks, PDLMs improved protein-DNA pair classification relative to independent-encoder baselines and supported prediction of mutation-associated binding changes. Zero-shot scoring recovered known changes in DNA-recognition specificity and prioritized variants with higher editing activity. Learned representations also supported cross-study prediction of base editor activity. After fine-tuning on transcription factor binding data, PDLMs predicted DNA motifs directly from protein sequences. Together, PDLMs provide a framework for interaction-aware biomolecular language modeling, with potential applications in protein engineering and genome editing.
ER -