Yeti: A compact protein structure tokenizer for reconstruction and multi-modal generation

• URL: http://arxiv.org/abs/2605.09981v1
• Date: Mon, 11 May 2026 04:49:47 GMT
• Title: Yeti: A compact protein structure tokenizer for reconstruction and multi-modal generation
• Abstract summary: We introduce Yeti, a compact protein structure tokenizer based on lookup free quantization and trained end to end with a flow matching objective for multimodal learning.<n>Compared to existing models, Yeti generally achieves the best codebook utilization and token diversity, and second best reconstruction accuracy (with 10x fewer parameters than ESM3) on diverse datasets.<n>The resulting multimodal model generates plausible structures under unconditional cogeneration of protein sequence and structures, achieving comparable results to 10x larger models.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Multimodal models that jointly reason over protein sequences, structures, and function annotations within a unified representation hold immense potential for integrating multimodal data and generating new proteins with designed functional properties. To utilize transformer architectures, such models require a tokenizer that converts protein structure from continuous atomic coordinates into discrete representations suitable for scalable multimodal training. The quality of such models are fundamentally upper bounded by the fidelity and expressiveness of the underlying tokenized structure. However, existing tokenizers prioritize reconstruction over generative abilities. To address these gaps, we introduce Yeti, a simple and compact protein structure tokenizer based on lookup free quantization and trained end to end with a flow matching objective for multimodal learning. Compared to existing models, Yeti generally achieves the best codebook utilization and token diversity, and second best reconstruction accuracy (with 10x fewer parameters than ESM3) on diverse datasets. To validate Yeti's generative capability, we trained a compact multimodal model jointly over its structure tokens and amino acid sequence entirely from scratch, with no pretrained initialization. The resulting multimodal model generates plausible structures under unconditional cogeneration of protein sequence and structures, achieving comparable results to 10x larger models. Together, these results demonstrate that Yeti is a compact and expressive protein structure tokenizer suitable for training multimodal models that cogenerates highly plausible sequences and structures.

Related papers

• SimpleDesign: A Joint Model for Protein Sequence and Structure Codesign [53.93453139057735]
  We present SimpleDesign, an effective multi-modal protein design model trained directly in the data space.<n>We train SimpleDesign on over 2M sequence-structure pairs achieving strong performance across co-design and unconditional sequence/structure generation benchmarks.
  arXiv  Detail & Related papers  (2026-09-03T05:25:05Z)
• Scaling Atomistic Protein Binder Design with Generative Pretraining and Test-Time Compute [64.59043351161004]
  We propose Proteina-Complexa, a novel fully atomistic binder generation method unifying both paradigms.<n>Proteina-Complexa sets a new state of the art in computational binder design benchmarks.<n>We also demonstrate interface hydrogen bond optimization, fold class-guided binder generation, and extensions to small molecule targets and enzyme design tasks.
  arXiv  Detail & Related papers  (2026-03-30T01:54:03Z)
• Protein Autoregressive Modeling via Multiscale Structure Generation [51.92004892768298]
  We present protein autoregressive modeling (PAR), the first multi-scale autoregressive framework for protein backbone generation.<n>We adopt noisy context learning and scheduled sampling, enabling robust backbone generation.<n>On the unconditional generation benchmark, PAR effectively learns protein distributions and produces backbones of high design quality.
  arXiv  Detail & Related papers  (2026-02-04T18:59:49Z)
• HD-Prot: A Protein Language Model for Joint Sequence-Structure Modeling with Continuous Structure Tokens [33.896084921181476]
  HD-Prot is a hybrid diffusion protein language model that embeds a continuous-valued diffusion head atop a discrete pLM.<n>It achieves competitive performance in unconditional sequence-structure co-generation, motif-scaffolding, protein structure prediction, and inverse folding tasks.
  arXiv  Detail & Related papers  (2025-12-17T06:46:27Z)
• ProteinAE: Protein Diffusion Autoencoders for Structure Encoding [64.77182442408254]
  We introduce ProteinAE, a novel and streamlined protein diffusion autoencoder.<n>ProteinAE directly maps protein backbone coordinates from E(3) into a continuous, compact latent space.<n>We demonstrate that ProteinAE achieves state-of-the-art reconstruction quality, outperforming existing autoencoders.
  arXiv  Detail & Related papers  (2025-10-12T14:30:32Z)
• All-atom inverse protein folding through discrete flow matching [11.47007092758602]
  We present ADFLIP (All-atom Discrete FLow matching Inverse Protein folding), a generative model for designing protein sequences conditioned on all-atom structural contexts.<n>Our model achieves state-of-the-art performance in single-structure and multi-structure inverse folding tasks, demonstrating excellent potential for all-atom protein design.
  arXiv  Detail & Related papers  (2025-07-04T11:57:38Z)
• DPLM-2: A Multimodal Diffusion Protein Language Model [75.98083311705182]
  We introduce DPLM-2, a multimodal protein foundation model that extends discrete diffusion protein language model (DPLM) to accommodate both sequences and structures. DPLM-2 learns the joint distribution of sequence and structure, as well as their marginals and conditionals. Empirical evaluation shows that DPLM-2 can simultaneously generate highly compatible amino acid sequences and their corresponding 3D structures.
  arXiv  Detail & Related papers  (2024-10-17T17:20:24Z)
• Sequence-Augmented SE(3)-Flow Matching For Conditional Protein Backbone Generation [55.93511121486321]
  We introduce FoldFlow-2, a novel sequence-conditioned flow matching model for protein structure generation.<n>We train FoldFlow-2 at scale on a new dataset that is an order of magnitude larger than PDB datasets of prior works.<n>We empirically observe that FoldFlow-2 outperforms previous state-of-the-art protein structure-based generative models.
  arXiv  Detail & Related papers  (2024-05-30T17:53:50Z)
• Learning the Language of Protein Structure [8.364087723533537]
  We introduce an approach using a vector-quantized autoencoder that effectively tokenizes protein structures into discrete representations.<n>To demonstrate the efficacy of our learned representations, we show that a simple GPT model trained on our codebooks can generate novel, diverse, and designable protein structures.
  arXiv  Detail & Related papers  (2024-05-24T16:03:47Z)
• EigenFold: Generative Protein Structure Prediction with Diffusion Models [10.24107243529341]
  EigenFold is a diffusion generative modeling framework for sampling a distribution of structures from a given protein sequence. On recent CAMEO targets, EigenFold achieves a median TMScore of 0.84, while providing a more comprehensive picture of model uncertainty.
  arXiv  Detail & Related papers  (2023-04-05T02:46:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.