Training data for miniML

This dataset contains annotated training data used to train the deep learning models used for miniML, a synaptic event detection method. Details about the method are outlined in the following publication:

O’Neill et al. A deep learning framework for automated and generalized synaptic event analysis. https://doi.org/10.7554/eLife.98485

The main miniML model was trained using data from the following synaptic recordings:

-              1_GC_mepsc_train.h5: Mouse cerebellar mossy fiber to granule cell (mEPSCs)

Training datasets for transfer learning for following types of recordings are provided:

-              2_GC_mepsp_train.h5: Mouse cerebellar mossy fiber to granule cell (mEPSPs)

-              3_NMJ_wt_train.h5: Drosophila neuromuscular junction WT (mEPSCs)

-              4_NMJ_glur2a_train.h5: Drosophila neuromuscular junction GluRIIA (mEPSCs)

-              5_ZF_sepsc_train.h5: Zebrafish telencephalon principal neurons (sEPSCs)

-              6_iglusnfr_train.h5: Cultured rat hippocampal neurons using iGluSnFR3 (glutamate transients)

-              7_asap5_train.h5: Cultured rat hippocampal neurons using ASAP5 (optical mEPSPs)

For more information on the training data, please refer to the accompanying readme.pdf file.

本数据集包含用于训练突触事件检测方法miniML所用深度学习模型的标注训练数据。有关该方法的详细细节，请参见下述发表论文：O’Neill等人的《用于自动化与通用化突触事件分析的深度学习框架》，DOI: 10.7554/eLife.98485。

miniML主模型使用下述突触记录数据完成训练：
- 1_GC_mepsc_train.h5：小鼠小脑苔藓纤维至颗粒细胞的微小兴奋性突触后电流（miniature excitatory postsynaptic currents, mEPSCs）

针对下述记录类型，提供了用于迁移学习的训练数据集：
- 2_GC_mepsp_train.h5：小鼠小脑苔藓纤维至颗粒细胞的微小兴奋性突触后电位（miniature excitatory postsynaptic potentials, mEPSPs）
- 3_NMJ_wt_train.h5：野生型（wild-type, WT）果蝇神经肌肉接头（neuromuscular junction, NMJ）的微小兴奋性突触后电流（mEPSCs）
- 4_NMJ_glur2a_train.h5：果蝇神经肌肉接头（neuromuscular junction, NMJ）GluRIIA型的微小兴奋性突触后电流（mEPSCs）
- 5_ZF_sepsc_train.h5：斑马鱼端脑主神经元的自发性兴奋性突触后电流（spontaneous excitatory postsynaptic currents, sEPSCs）
- 6_iglusnfr_train.h5：使用iGluSnFR3标记的大鼠体外培养海马神经元（谷氨酸瞬态信号）
- 7_asap5_train.h5：使用ASAP5标记的大鼠体外培养海马神经元（光学成像微小兴奋性突触后电位）

如需获取更多训练数据相关信息，请参阅附带的readme.pdf文件。