kitAb: an automated framework for assay-specific antibody developability modeling and large-scale landscape analysis
- 1MSU Institute for Artificial Intelligence, Lomonosov Moscow State University, Moscow, Russia
Abstract
Antibody developability is determined by multiple physicochemical properties, and the relevance of individual properties varies among experimental assays. Existing computational approaches, however, often provide fixed assessments or rely on computationally intensive descriptors. We present kitAb, an open-source framework for scalable physicochemical characterization of antibody Fv regions and assay-specific developability modeling. kitAb calculates 59 interpretable sequence- and structure-based descriptors and integrates them with automated feature selection and regression. We evaluated kitAb across ten public datasets comprising 46 experimental endpoints, using sequence-aware or predefined splits where feasible. Predictive performance was strongly assay-dependent. In nested evaluations with sequence-aware or predefined data splits, models using kitAb descriptors achieved predictive performance comparable on average to those using PROPERMAB descriptors, while outperforming models based on TAP or sequence-only features. In a single-structure CPU timing benchmark, kitAb descriptor calculation was approximately 11-fold faster than PROPERMAB. Predictive performance was similar across the three structure-generation backends, and broader AutoML searches did not substantially improve mean performance. We further applied the same descriptor representation to approximately 3.4 million paired natural antibodies and 706 Phase II or later therapeutic antibodies. Repertoire-scale analysis revealed systematic differences in surface charge, hydrophobic and aromatic organization, CDR length and germline-associated physicochemical profiles. The kitAb code and analysis workflows are openly available at https://github.com/katyachemistry/kitAb/.
Related
- An interpretable open platform for sequence-based antibody developability prediction
- Near real-time data on the human neutralizing antibody landscape to influenza virus in summer of 2026 shows antigenic advance of H3N2 subclade K region D mutants and H1N1 D.3.1.1 Sa mutants
- Systemic antibody repertoires target specific microbial antigens in patients with colorectal cancer
- Single-Cell Transcriptomic Landscape of the Gill in Hypoxia-Stressed Tachysurus vachellii (Richardson, 1846): Insights into Cell-Type-Specific Responses of Different Epithelial Cells and Erythrocytes
- 895 Engineering GPRC5D multispecific antibodies: parallel engineering to optimize affinity, developability, and IP
BibTeX
@misc{trofimova2026kitab,
title = {{kitAb: an automated framework for assay-specific antibody developability modeling and large-scale landscape analysis}},
author = {E. Trofimova and E. Shaburova and D. Antonets and V. Ramensky},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.09.750540},
url = {https://www.biorxiv.org/content/10.64898/2026.10.09.750540v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - kitAb: an automated framework for assay-specific antibody developability modeling and large-scale landscape analysis
AU - E. Trofimova
AU - E. Shaburova
AU - D. Antonets
AU - V. Ramensky
PY - 2026
DA - 2026/10/09
PB - bioRxiv
DO - 10.64898/2026.10.09.750540
UR - https://www.biorxiv.org/content/10.64898/2026.10.09.750540v1
AB - Antibody developability is determined by multiple physicochemical properties, and the relevance of individual properties varies among experimental assays. Existing computational approaches, however, often provide fixed assessments or rely on computationally intensive descriptors. We present kitAb, an open-source framework for scalable physicochemical characterization of antibody Fv regions and assay-specific developability modeling. kitAb calculates 59 interpretable sequence- and structure-based descriptors and integrates them with automated feature selection and regression. We evaluated kitAb across ten public datasets comprising 46 experimental endpoints, using sequence-aware or predefined splits where feasible. Predictive performance was strongly assay-dependent. In nested evaluations with sequence-aware or predefined data splits, models using kitAb descriptors achieved predictive performance comparable on average to those using PROPERMAB descriptors, while outperforming models based on TAP or sequence-only features. In a single-structure CPU timing benchmark, kitAb descriptor calculation was approximately 11-fold faster than PROPERMAB. Predictive performance was similar across the three structure-generation backends, and broader AutoML searches did not substantially improve mean performance. We further applied the same descriptor representation to approximately 3.4 million paired natural antibodies and 706 Phase II or later therapeutic antibodies. Repertoire-scale analysis revealed systematic differences in surface charge, hydrophobic and aromatic organization, CDR length and germline-associated physicochemical profiles. The kitAb code and analysis workflows are openly available at https://github.com/katyachemistry/kitAb/.
ER -