Moreza009/AAV_datasets

AAVGen: Precision Engineering of Adeno-associated Virus for Renal Selective Targeting

---

Overview

This is the curated and processed dataset used to train AAVGen, a generative AI framework for de novo design of adeno-associated virus (AAV) capsids with enhanced multi-trait profiles. The dataset aggregates experimental fitness measurements for AAV2 and AAV9 capsid variants across multiple functional properties, including production efficiency, kidney tropism, and thermostability.

The dataset contains 820,993 total examples (with repetitive sequences in different splits) across 9 splits, covering two AAV serotypes (AAV2 and AAV9).

---

Abstract

Adeno-associated viruses (AAVs) are promising vectors for gene therapy, but their native serotypes face limitations in tissue tropism, immune evasion, and production efficiency. Engineering capsids to overcome these hurdles is challenging due to the vast sequence space and the difficulty of simultaneously optimizing multiple functional properties. The complexity also adds when it comes to the kidney, which presents unique anatomical barriers and cellular targets that require precise and efficient vector engineering. Here, we present AAVGen, a generative artificial intelligence framework for de novo design of AAV capsids with enhanced multi-trait profiles. AAVGen integrates a protein language model (PLM) with supervised fine-tuning (SFT) and a reinforcement learning technique termed Group Sequence Policy Optimization (GSPO). The model is guided by a composite reward signal derived from three ESM-2-based regression predictors, each trained to predict a key property: production fitness, kidney tropism, and thermostability. Our results demonstrate that AAVGen produces a diverse library of novel VP1 protein sequences. In silico validations revealed that the majority of the generated variants have superior performance across all three employed indices, indicating successful multi-objective optimization. Furthermore, structural analysis via AlphaFold3 confirms that the generated sequences preserve the canonical capsid folding despite sequence diversification. AAVGen establishes a foundation for data-driven viral vector engineering, accelerating the development of next-generation AAV vectors with tailored functional characteristics.

---

Dataset Structure

Fields

Field	Type	Description
final_seq	string	VP1 capsid protein amino acid sequence
fitness_score	float64	Experimentally measured scores for the given assay
aav_type	string	AAV serotype identifier (e.g., AAV2, AAV9)

---

Splits

The dataset is divided into 9 splits organized by serotype and assay type:

AAV2 — sourced from Ogden et al. and Bryant et al.

Split	Description
AAV2_Thermostability	Thermostability fitness scores for AAV2 variants
AAV2_Kidney_Tropism	Kidney tropism fitness scores for AAV2 variants
AAV2_production_main_merged_final	Production efficiency fitness scores for AAV2 variants (merged from multiple experimental replicates)

AAV9 — sourced from Eid et al.

Split	Description
AAV9_THLE_tr	Transduction efficiency in THLE-2 (normal liver) cells
AAV9_HepG2_bind	Binding efficiency in HepG2 (hepatocellular carcinoma) cells
AAV9_HepG2_tr	Transduction efficiency in HepG2 cells
AAV9_Liver	In vivo liver tropism fitness scores
AAV9_Production	Production efficiency fitness scores for AAV9 variants
AAV9_THLE_bind	Binding efficiency in THLE-2 cells

---

Usage

from datasets import load_dataset

# Load a specific split
ds = load_dataset("mohammad-gh009/AAVGen", split="AAV2_Kidney_Tropism")

# Load all splits
ds = load_dataset("mohammad-gh009/AAVGen")
print(ds)

---

Source Studies

This dataset aggregates and processes data from the following published studies:

1. Ogden et al. — Comprehensive AAV capsid fitness landscape via deep mutational scanning. Science, 2019.
2. Bryant et al. — Deep diversification of an AAV capsid protein by machine learning. Nature Biotechnology, 2021.
3. Eid et al. — In vivo evolution of AAV capsids by massively parallel sequencing and selection. Nature Biotechnology, 2022.

---

Citation

If you use this dataset, please cite the AAVGen paper:

@misc{ghaffarzadehesfahani2026aavgenprecisionengineeringadenoassociated,
      title={AAVGen: Precision Engineering of Adeno-associated Viral Capsids for Renal Selective Targeting}, 
      author={Mohammadreza Ghaffarzadeh-Esfahani and Yousof Gheisari},
      year={2026},
      eprint={2602.18915},
      archivePrefix={arXiv},
      primaryClass={q-bio.QM},
      url={https://arxiv.org/abs/2602.18915}, 
}

---

License

This dataset is released under the Apache 2.0 License.