TREM2结合物设计数据集
包含TREM2蛋白结合物设计序列、结合强度与结构预测数据，用于AI代理设计效果比较，与阿尔茨海默病相关。
基本信息
资源简介
TREM2结合物设计数据集
【数据内容】
源站原始描述：Comparison of Muni and Anthropic agentic TREM2 binder design results - agitter/agentic-TREM2-binders
【适用方向】
组学数据适用于生物标志物发现与分子分型相关建模研究。
【获取与许可】
源站页面：https://github.com/agitter/agentic-TREM2-binders
下载信息
注册下载
需要注册 Kaggle 账号并登录后下载，适合需要跟踪下载记录和使用 API 的用户。
暂未开放公开下载
无需注册即可直接获取公开样本或文档，适合快速预览和评估数据集质量。
免登录有偿下载
公开数据集受托下载与技术交付服务。
提供高速下载与技术交付服务（收技术服务费，非数据销售）
暂未开放当前数据集主要来源为 Kaggle 公开托管，完整影像包建议通过原始链接或 Kaggle API 获取。
使用方式
数据集获取
git clone https://github.com/agitter/agentic-TREM2-binders.git
curl -L -o repo.zip https://github.com/agitter/agentic-TREM2-binders/archive/refs/heads/main.zip
unzip repo.zip
源站 README 摘录（使用方式）
Comparison of Muni and Anthropic agentic TREM2 binder design results
This repository compares AI agents’’ designs of TREM2 protein binders that were recently shared by Muni and Anthropic.
All artifacts except those in /data are from Claude Opus 5.
The code may not run locally, and external software dependencies are not provided.
METHODS_ARCHIVE.md provides an overview of the analyses, limitations, and software Claude used.
Workflow overview
STAGE 1 two collections (100 designs) artifacts unsuffixed, figures unchanged
=======================================================================================
designs.fasta filter target=TREM2> 90 seqs, 3 workflows -+
design_summary.csv binder_final> 72 bind / 18 not -+
+> 100 designs
Muni CSV (utf-8-sig) binding_strength> 9 Strong / 1 null -+ (null = expression
muni_files.zip 10 Boltz2 complexes-+ failure, a measured
negative)
TREM2.zip predicted_boltz2_1to1.cif> chain matched BY SEQUENCE > 100 binder PDBs
(chain B here, chain A in designed.cif)
|
+> Foldseek TM-align, exhaustive > 100x100 TM max-symmetrized> tm_all.npy
| \-> CIGAR backtrace > superposition
|
\> Biopython BLOSUM62 semi-global > 100x100 % identity > sid_all.npy
(cross-checked vs MMseqs2)
tm_all > Louvain modules > community-seeded force layout > node positions (FROZEN)
\> edges at TM >= 0.80 -+
sid_all \> edges at identity >= 40% -+> same positions, two edge layers
structures > superpose onto global medoid > py2Dmol best_view (ONE shared frame)
> tube render, hue = workflow > node bitmaps
|
binders x2 panels <-+-> non-binders x2 panels
\-> combined x2 panels, hue = binder status
STAGE 2 add the hackathon (200 designs)
## 数据加载示例（表格/文本类）
```python
import pandas as pd, glob, os
files = (glob.glob(os.path.join(path, "**", "*.csv"), recursive=True)
+ glob.glob(os.path.join(path, "**", "*.tsv"), recursive=True)
+ glob.glob(os.path.join(path, "**", "*.xlsx"), recursive=True))
print("数据文件:", files)
df = pd.read_csv(files[0])
print(df.shape); print(df.columns.tolist()); print(df.head(3))
数据缺失？
依托陕西、四川两大基地，我们与超过十家三甲医院建立直接合作关系，覆盖合规授权、采集治理、专业标注、数据交付的全流程，为AI医疗团队提供即拿即用的高质量临床数据。
⚡️ 需要数据支持或标注服务？立即联系我们获取专业方案。
