遇见数据集

SpikeLab: Agentic tools for spike data analysis - Supplementary Data

收藏
Zenodo2026-04-26 更新2026-05-26 收录
官方服务:

资源简介:

This directory contains the data and resources accompanying the manuscript: "SpikeLab: Agentic tools for spike data analysis". Three components are included: the distributable SpikeLab skill, the complete benchmark data underlying the model comparison, and the spike-sorted recordings used across all analyses. Supplementary Data 1: SpikeLab Skill The distributable SpikeLab skill. Contains `SKILL.md` (the entry-point skill that handles installation, environment setup, and delegation to specialized analysis, sorting, education, and development roles) and `INSTALL.md` (installation guide for the SpikeLab library). Skill files are plain markdown documents that can be used with any agentic coding tool that supports custom instructions — simply provide the skill file as context. Supplementary Data 2: Benchmark Complete raw outputs and processed data from the controlled benchmark used in Fig. 2. Contains full session transcripts, generated analysis scripts, output figures, and quantitative scoring data for all 9 benchmark runs (3 conditions x 3 replicates). See `benchmark/README.md` for detailed structure and run identifiers. Supplementary Data 3: Sorted Recordings Spike-sorted recording data used throughout the manuscript, stored as serialized SpikeData objects (`.pkl`). Includes organoid recordings at five developmental timepoints (D0, D3, D10, D30, D50) and in vivo recordings from mouse and human cortex. These files are the direct inputs to the analysis scripts and figure code.

本目录包含伴随学术论文《SpikeLab:用于尖峰信号数据分析的AI智能体(AI Agent)工具》的数据集与配套资源。本次公开内容涵盖三个组成部分:可分发的SpikeLab技能、支撑模型对比实验的完整基准数据集,以及全分析流程所用的尖峰分选记录数据。 ### 补充数据1:SpikeLab技能 可分发的SpikeLab技能包。内含`SKILL.md`(作为入口技能文件,负责处理安装、环境配置,并将任务委派至专业化的分析、分选、教学与开发角色)与`INSTALL.md`(SpikeLab库安装指南)。本技能文件均为纯Markdown文档,可直接用于任意支持自定义指令的AI智能体(AI Agent)编码工具——仅需将该技能文件作为上下文输入即可。 ### 补充数据2:基准数据集 本部分包含论文中图2所用受控基准实验的完整原始输出与处理后数据,涵盖全部9次基准测试(3种实验条件×3次重复)的完整会话转录文本、生成的分析脚本、输出图表与量化评分数据。详细的目录结构与运行标识符说明请参见`benchmark/README.md`。 ### 补充数据3:分选记录数据 本部分包含论文全流程使用的尖峰分选记录数据,以序列化SpikeData对象格式存储(文件后缀为`.pkl`)。其中包含5个发育时间点(D0、D3、D10、D30、D50)的类器官记录数据,以及小鼠与人类皮层的在体记录数据。这些文件为分析脚本与图表生成代码的直接输入源。

提供机构:
Zenodo
创建时间:
2026-04-26
二维码
社区交流群
二维码
科研交流群
商业服务