---
title: "Tabular in-context learning with coarse composition for multi-activity antimicrobial peptide profi"
id: "biorxiv-4-coarse-composition-suffices-tabular-in-context-learning-for-multi-activity"
canonical_url: "https://medichelpline.com/clinical-feed/biorxiv-4-coarse-composition-suffices-tabular-in-context-learning-for-multi-activity"
content_type: "clinical_feed_article"
specialty: "Pharmacology"
source_name: "bioRxiv (Biomedical Preprints)"
source_url: "https://www.biorxiv.org/content/10.64898/2026.08.27.747591v1?rss=1"
published_at: "2026-08-28T12:00:00.000Z"
evidence_level: "Verified Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Tabular in-context learning with coarse composition for multi-activity antimicrobial peptide profi
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/biorxiv-4-coarse-composition-suffices-tabular-in-context-learning-for-multi-activity
- **Specialty:** [Pharmacology](https://medichelpline.com/clinical-feed/pharmacology.md)
- **Primary Source:** bioRxiv (Biomedical Preprints)
- **Source URL:** [Original Journal Publication](https://www.biorxiv.org/content/10.64898/2026.08.27.747591v1?rss=1)
- **Published At:** 2026-08-28T12:00:00.000Z
- **Evidence Rating:** Verified Feed
## Executive GIST (TL;DR)
- The study evaluates multi-label activity prediction for **antimicrobial peptides (AMPs)**, arguing this reflects real screening needs better than binary antimicrobial classification. - The ESCAPE benchmark (82,359 peptides; five labels) is used as the evaluation dataset for multi-activity prediction. - Prior top methods used multimodal, structure-conditioned deep models that require expensive training and tuning. - The authors present a simple, sequence-only pipeline built from 330 interpretable sequence descriptors combined with **TabPFN**, a tabular foundation model that performs in-context prediction without gradient-based training or hyperparameter optimization. - A label-powerset TabPFN model achieves **mAP-5 = 77.8%**, improving on the previously reported best of 72.1% on ESCAPE. - A probabilistic classifier chain approach matches or exceeds the best published average precision for each of the five labels simultaneously. - Performance gains remain under a single-fold training protocol, indicating results are not solely due to larger training sets. - Gains are especially pronounced for remote homologues (more than 11.2 percentage points improvement for sequences below 30% identity). - Ablation experiments show predicted structure is unnecessary at inference and that no single descriptor family drives performance; ten global physicochemical scalar features recover 91% of the full-feature performance. - Explicitly modelling label dependence provides benefits for scarce activities and enables prioritization of which activity to assay next given partial positive evidence. - The authors declare no competing interests.
## Clinical Analysis & Structured Key Points
Coarse composition suffices: tabular in-context learning for multi-activity antimicrobial peptide profiling | bioRxiv Skip to main content New Results Coarse composition suffices: tabular in-context learning for multi-activity antimicrobial peptide profiling Raunak Kumar , Anuj Pal , Dhruvi Solanki , View ORCID Profile Parikshit Pareek , View ORCID Profile Juhi Singh , View ORCID Profile Jitin Singla doi: https://doi.org/10.64898/2026.08.27.747591 Raunak Kumar 1 IIT Roorkee; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Anuj Pal 1 IIT Roorkee; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Dhruvi Solanki 2 IISc Bangalore Find this author on Google Scholar Find this author on PubMed Search for this author on this site Parikshit Pareek 1 IIT Roorkee; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Parikshit Pareek Juhi Singh 2 IISc Bangalore Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Juhi Singh Jitin Singla 1 IIT Roorkee; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Jitin Singla For correspondence: jsingla{at}bt.iitr.ac.in Abstract Info/History Metrics Preview PDF Abstract Antimicrobial peptides (AMPs) often act against multiple pathogen classes, making multi-label activity prediction a more realistic screening target than binary antimicrobial classification. The ESCAPE benchmark formalizes this setting, but leading approaches typically rely on multimodal, structure-conditioned deep models that are costly to train and tune. We show that a simple, sequence-only pipeline can match and surpass these methods by combining 330 interpretable sequence descriptors with TabPFN, a tabular foundation model that performs in-context prediction in a single forward pass without gradient-based training or hyperparameter search. On ESCAPE (82,359 peptides; five labels), a label-powerset TabPFN model achieves mAP-5=77.8%, improving on the previously best reported 72.1%. A probabilistic classifier chain is the first method to match or exceed the best published average precision on each of the five labels simultaneously. The gains persist under the prior state-of-the-art single-fold training protocol, indicating they are not a training-set-size artefact, and are largest for remote homologues (+11.2 points below 30% sequence identity). Ablations further show that predicted structure is unnecessary at inference and that performance is not driven by any single descriptor family: ten global physicochemical scalars recover 91% of full-feature performance. Finally, explicitly modelling label dependence yields targeted benefits for scarce activities and supports ranking which activity to assay next from partial positive evidence. Competing Interest Statement The authors have declared no competing interest. Copyright The copyright holder for this preprint is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY 4.0 International license . Back to top Previous Next Posted August 28, 2026. Download PDF Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Coarse composition suffices: tabular in-context learning for multi-activity antimicrobial peptide profiling Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Coarse composition suffices: tabular in-context learning for multi-activity antimicrobial peptide profiling Raunak Kumar , Anuj Pal , Dhruvi Solanki , Parikshit Pareek , Juhi Singh , Jitin Singla bioRxiv 2026.08.27.747591; doi: https://doi.org/10.64898/2026.08.27.747591 Share This Article: Copy Citation Tools Coarse composition suffices: tabular in-context learning for multi-activity antimicrobial peptide profiling Raunak Kumar , Anuj Pal , Dhruvi Solanki , Parikshit Pareek , Juhi Singh , Jitin Singla bioRxiv 2026.08.27.747591; doi: https://doi.org/10.64898/2026.08.27.747591 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Bioinformatics Subject Areas All Articles Animal Behavior and Cognition (7945) Biochemistry (18565) Bioengineering (14718) Bioinformatics (43981) Biophysics (22375) Cancer Biology (19499) Cell Biology (26670) Clinical Trials (138) Developmental Biology (13855) Ecology (20805) Epidemiology (2067) Evolutionary Biology (25250) Genetics (16067) Genomics (23351) Immunology (18528) Microbiology (42092) Molecular Biology (17885) Neuroscience (92579) Paleontology (691) Pathology (2958) Pharmacology and Toxicology (5051) Physiology (8029) Plant Biology (15846) Scientific Communication and Education (2090) Synthetic Biology (4523) Systems Biology (10156) Zoology (2368)
## Related Clinical Research

- [Antimicrobial Stewardship Challenges in Low- and Middle-Income Countries](https://medichelpline.com/clinical-feed/plos-medicine-0-antimicrobial-stewardship-in-low-and-middle-income-countries-access-versus.md)
- [Knowledge, Attitudes, and Behaviour toward Japanese Encephalitis Vaccination: A Systematic Review](https://medichelpline.com/clinical-feed/pubmed-42556299.md) (DOI: 10.1016/j.vaccine.2026.128999)
- [Geospatial methods for mapping zero-dose and routine childhood immunisation coverage in low-income](https://medichelpline.com/clinical-feed/pubmed-42442083.md) (DOI: 10.1016/j.vaccine.2026.128904)
- [ML‑Enabled Multiscale QSP–PBPK Platform to Support Development of mRNA Therapeutics](https://medichelpline.com/clinical-feed/medrxiv-16-towards-a-ml-powered-multiscale-computational-platform-based-on-qsp-and-pbpk.md)
- [Cassidy report faults FDA oversight of mifepristone manufacturers](https://medichelpline.com/clinical-feed/stat-news-5-stat-cassidy-criticizes-fda-oversight-of-abortion-pill-manufacturers-in-new.md)

## Navigation
- [← Back to Pharmacology Feed](https://medichelpline.com/clinical-feed/pharmacology.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.