An official website of the United States government
Signal peptides (SPs) play a crucial role in protein translocation in cells. The development of large protein language models (PLMs) and prompt-based learning provide a new opportunity for SP prediction, especially for the categories with limited annotated data. We present a parameter-efficient fine-tuning (PEFT) framework for SP prediction, PEFT-SP, to effectively utilize pretrained PLMs. We integrated low-rank adaptation (LoRA) into ESM-2 models to better leverage the protein sequence evolutionary knowledge of PLMs. Experiments show that PEFT-SP using LoRA enhances state-of-the-art results, leading to a maximum Matthews correlation coefficient (MCC) gain of 87.3% for SPs with small training samples and an overall MCC gain of 6.1%. Furthermore, we also employed two other PEFT methods, prompt tuning and adapter tuning, in ESM-2 for SP prediction. More elaborate experiments show that PEFT-SP using adapter tuning can also improve the state-of-the-art results by up to 28.1% MCC gain for SPs with small training samples and an overall MCC gain of 3.8%. LoRA requires fewer computing resources and less memory than the adapter tuning during the training stage, making it possible to adapt larger and more powerful protein models for SP prediction.
more »
« less
This content will become publicly available on November 5, 2025
Protein language models learn evolutionary statistics of interacting sequence motifs
Protein language models (pLMs) have emerged as potent tools for predicting and designing protein structure and function, and the degree to which these models fundamentally understand the inherent biophysics of protein structure stands as an open question. Motivated by a finding that pLM-based structure predictors erroneously predict nonphysical structures for protein isoforms, we investigated the nature of sequence context needed for contact predictions in the pLM Evolutionary Scale Modeling (ESM-2). We demonstrate by use of a “categorical Jacobian” calculation that ESM-2 stores statistics of coevolving residues, analogously to simpler modeling approaches like Markov Random Fields and Multivariate Gaussian models. We further investigated how ESM-2 “stores” information needed to predict contacts by comparing sequence masking strategies, and found that providing local windows of sequence information allowed ESM-2 to best recover predicted contacts. This suggests that pLMs predict contacts by storing motifs of pairwise contacts. Our investigation highlights the limitations of current pLMs and underscores the importance of understanding the underlying mechanisms of these models.
more »
« less
- Award ID(s):
- 2032259
- PAR ID:
- 10631241
- Publisher / Repository:
- PNAS
- Date Published:
- Journal Name:
- Proceedings of the National Academy of Sciences
- Volume:
- 121
- Issue:
- 45
- ISSN:
- 0027-8424
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Protein language models (pLMs) trained on a large corpus of protein sequences have shown unprecedented scalability and broad generalizability in a wide range of predictive modeling tasks, but their power has not yet been harnessed for predicting protein–nucleic acid binding sites, critical for characterizing the interactions between proteins and nucleic acids. Here, we present EquiPNAS, a new pLM-informed E(3) equivariant deep graph neural network framework for improved protein–nucleic acid binding site prediction. By combining the strengths of pLM and symmetry-aware deep graph learning, EquiPNAS consistently outperforms the state-of-the-art methods for both protein–DNA and protein–RNA binding site prediction on multiple datasets across a diverse set of predictive modeling scenarios ranging from using experimental input to AlphaFold2 predictions. Our ablation study reveals that the pLM embeddings used in EquiPNAS are sufficiently powerful to dramatically reduce the dependence on the availability of evolutionary information without compromising on accuracy, and that the symmetry-aware nature of the E(3) equivariant graph-based neural architecture offers remarkable robustness and performance resilience. EquiPNAS is freely available at https://github.com/Bhattacharya-Lab/EquiPNAS.more » « less
-
Abstract Protein language models (pLMs) have been widely adopted for various protein and peptide-related downstream tasks and demonstrated promising performance. However, short peptides are significantly underrepresented in commonly used pLM training datasets. For example, only 2.8% of sequences in the UniProt Reference Cluster (UniRef) contain fewer than 50 residues, which potentially limits the effectiveness of pLMs for peptide-specific applications. Here, we present PepBERT, a lightweight and efficient peptide language model specifically designed for encoding peptide sequences. Two versions of the model—PepBERT-large (4.9 million parameters) and PepBERT-small (1.86 million parameters)—were pretrained from scratch using four custom peptide datasets and evaluated on nine peptide-related downstream prediction tasks. Both PepBERT models achieved performance superior to or comparable to the benchmark model, ESM-2 with 7.5 million parameters, on 8 out of 9 datasets. Overall, PepBERT provides a compact yet effective solution for generating high-quality peptide representations for downstream applications. By enabling more accurate representation and prediction of bioactive peptides, PepBERT can accelerate the discovery of food-derived bioactive peptides with health-promoting properties, supporting the development of sustainable functional foods and value-added utilization of food processing by-products. The datasets, source codes, pretrained models, and tutorials for the usage of PepBERT are available athttps://github.com/dzjxzyd/PepBERT.more » « less
-
Accurate detection of protein sequence homology is essential for understanding evolutionary relationships and predicting protein functions, particularly for detecting remote homology in the “twilight zone” (20-35% sequence similarity), where traditional sequence alignment methods often fail. Recent studies show that embeddings from protein language models (pLM) can improve remote homology detection over traditional methods. Alignment-based approaches, such as those combining pLMs with dynamic programming alignment, further improve performance but often suffer from noise in the resulting similarity matrices. To address this, we evaluate a newly developed embedding-based sequence alignment approach that refines residue-level embedding similarity using K-means clustering and double dynamic programming (DDP). We show that the incorporation of clustering and DDP contributes substantially to the improved performance in detecting remote homology. Experimental results demonstrate that our approach outperforms both traditional and state-of-the-art approaches based on pLMs on several benchmarks. Our study illustrates embedding-based alignment refined with clustering and DDP offers a powerful approach for identifying remote homology, with potential to evolve further as pLMs continue to advance.more » « less
-
Abstract MotivationRecent advancements in natural language processing have highlighted the effectiveness of global contextualized representations from protein language models (pLMs) in numerous downstream tasks. Nonetheless, strategies to encode the site-of-interest leveraging pLMs for per-residue prediction tasks, such as crotonylation (Kcr) prediction, remain largely uncharted. ResultsHerein, we adopt a range of approaches for utilizing pLMs by experimenting with different input sequence types (full-length protein sequence versus window sequence), assessing the implications of utilizing per-residue embedding of the site-of-interest as well as embeddings of window residues centered around it. Building upon these insights, we developed a novel residual ConvBiLSTM network designed to process window-level embeddings of the site-of-interest generated by the ProtT5-XL-UniRef50 pLM using full-length sequences as input. This model, termed T5ResConvBiLSTM, surpasses existing state-of-the-art Kcr predictors in performance across three diverse datasets. To validate our approach of utilizing full sequence-based window-level embeddings, we also delved into the interpretability of ProtT5-derived embedding tensors in two ways: firstly, by scrutinizing the attention weights obtained from the transformer’s encoder block; and secondly, by computing SHAP values for these tensors, providing a model-agnostic interpretation of the prediction results. Additionally, we enhance the latent representation of ProtT5 by incorporating two additional local representations, one derived from amino acid properties and the other from supervised embedding layer, through an intermediate fusion stacked generalization approach, using an n-mer window sequence (or, peptide/fragment). The resultant stacked model, dubbed LMCrot, exhibits a more pronounced improvement in predictive performance across the tested datasets. Availability and implementationLMCrot is publicly available at https://github.com/KCLabMTU/LMCrot.more » « less
