| --- |
| license: mit |
| task_categories: |
| - text-generation |
| language: |
| - en |
| tags: |
| - biology |
| - antibodies |
| size_categories: |
| - 100M<n<1B |
| --- |
| |
| # CoSiNE Training Dataset |
|
|
| Training data for [CoSiNE](https://github.com/thematrixmaster/cosine) model, which simulates antibody affinity maturation. Please read our [paper](https://arxiv.org/abs/2602.18982) for more details! |
|
|
| ## Dataset Summary |
|
|
| This dataset consists of approximately **2 million B-cell receptor (BCR) parent-child sequence transitions** derived from **120,000 clonal families** across **555 individual donors**. |
| The data was processed using a rigorous phylogenetic inference pipeline to capture the nuances of somatic hypermutation: |
|
|
| 1. **Clonal Inference:** Sequences were clustered into clonal families and naive germlines were inferred using `partis`. |
| 2. **Quality Filtering:** We retained only productive sequences (no stop codons, conserved CDR3 anchors) and excluded sequences with mutations in conserved signature cysteines. |
| 3. **Phylogenetic Reconstruction:** Phylogenetic trees and ancestral sequences were inferred using **IQ-TREE** under a K80 substitution model. |
| 4. **Paired-Chain Modeling:** For paired heavy and light chain data, we utilized an edge-linked-proportional partition model to account for distinct evolutionary rates across chains. |
|
|
| The final training set consists of **Parent-Child Pairs (PCPs)** extracted from the edges of these phylogenetic trees, representing a comprehensive map of the evolutionary trajectories within the adaptive immune system. For a detailed description of the processing pipeline, please refer to https://elifesciences.org/reviewed-preprints/109644v1. |
|
|
| ## Dataset Sources |
|
|
| The dataset was compiled using B-cell receptor (BCR) sequencing datasets from five sources: |
|
|
| * [Jaffe-2022](https://www.nature.com/articles/s41586-022-05371-z) |
| * [Tang-2022](https://www.sciencedirect.com/science/article/pii/S2589004221016382) |
| * [Vergani-2017](https://pubmed.ncbi.nlm.nih.gov/28959265/) |
| * [Engelbrecht-2025](https://www.nature.com/articles/s41467-025-66759-9) |
| * [Rodriguez-2023](https://www.nature.com/articles/s41467-023-40070-x) |
|
|
| ## Citation |
|
|
| Consider citing our paper if you use CoSiNE in your research! |
|
|
| ```bibtex |
| @article{Lu2026ConditionallySN, |
| title={Conditionally Site-Independent Neural Evolution of Antibody Sequences}, |
| author={Stephen Zhewen Lu and Aakarsh Vermani and Kohei Sanno and Jiarui Lu and IV FrederickA.Matsen and Milind Jagota and Yun S. Song}, |
| journal={ArXiv}, |
| year={2026}, |
| url={https://api.semanticscholar.org/CorpusID:285973749} |
| } |
| ``` |
|
|
| ## Dataset Card Contact |
|
|
| For questions or issues, please contact: |
| - Stephen Z. Lu (stephen.lu@berkeley.edu) |
| - Aakarsh Vermani (aakarshv@berkeley.edu) |
|
|
