{
  "abstract": "Protein Language Models (PLMs) have made remarkable progress following scaling laws established in natural language processing across sequence- and structure-based tasks, yet the potential of tokenization remains underexploited. Unlike human language, proteins preserve structure despite extensive sequence variation a property standard tokenization strategies fundamentally fail to capture. We introduce ZEST (Zoned Encoding of Sequence Traits), an evolution-informed vocabulary derived from conserved regions of multiple sequence alignments. ZEST allows embedding domain-level biological priors directly at the tokenization stage rather than learning them implicitly through scale. ZEST natively compresses sequences to an average token length of 4 residues, enabling our model to process 4,000 residues within a standard 1024-token context window. Building on this, we present LEMON (Layered Extraction of Molecular Ordering from Nature), a compact 200M-parameter sequence-based model for detection of remote homology between protein sequences trained on a single H100 GPU for one week. Despite its modest size, LEMON outperforms state-of-the-art models ranging from 600M to 3B parameters. Our results demonstrate that evolution-informed tokenization can substitute for massive parameter scaling, opening a new direction for efficient, biologically-grounded protein representation learning. All code, model weights, and results are publicly available under the MIT license.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2609.37675",
  "author_total": 3,
  "authors": [
    {
      "name": "Biswajit Banerjee"
    },
    {
      "name": "Claudia Alvarez Carreno"
    },
    {
      "name": "Anton S. Petrov"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/3hwqqyrf6yyhb.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2609.37675",
  "fulltext_url": null,
  "id": "3hwqqyrf6yyhb",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2609.37675",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "q-bio.BM"
    ],
    [
      "openalex",
      "1312|Machine Learning in Bioinformatics"
    ]
  ],
  "openalex": "W7214916128",
  "pdf_url": "https://arxiv.org/pdf/2609.37675",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-29",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/3hwqqyrf6yyhb.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:25:36Z",
      "native_id": "2609.37675",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2609.37675"
    }
  ],
  "subjects": [
    {
      "field": "biochemistry",
      "origin": "arxiv:q-bio.BM",
      "subfield": "biophysics"
    }
  ],
  "title": "LEMON-ZEST: Evolution-Informed Tokenization for Efficient Protein Language Modeling",
  "title_text": "LEMON-ZEST: Evolution-Informed Tokenization for Efficient Protein Language Modeling",
  "type": "preprint",
  "updated_at": "2026-10-30T12:25:36Z",
  "url": "https://research.draconitic.com/paper/3hwqqyrf6yyhb",
  "venue": "arXiv",
  "versions": []
}