遇见数据集

Dataset and Model for "ReShuffle-MS"

收藏
Zenodo2025-12-15 更新2026-05-26 收录
官方服务:

资源简介:

This repository contains the dataset and models for the research presented in the paper "ReShuffle-MS". 'data.zip': Escherichia coli data from Shanghai Sixth People's Hospital.(1) The MALDI-TOF mass spectra in the form of '.txt' files, in the 'raw' folder. The name of each txt file is the 'SPEC_NUM'. The first column is the mass-to-charge ratio, and the second column is the corresponding intensity.(2)'preprocessed' all spectra after the application of the preprocessing pipeline and binning along the mass-to-charge-ratio axis with a bin size of 3Da, resulting in 6000 feature bins. The dataset was split into training, validation, and test sets : 20% of the data were isolated as a final, held-out test set using stratified sampling, and the remaining 80% were further divided into a training set (90%, or 72% of the total data) and a validation set (10%, or 8% of the total).(3)'202403-11_E.coli.csv' shows the results of the antibiotic susceptibility testing (AST) of this batch of Escherichia coli. 'model.zip': Contains the seven classifier models used in our paper, each of which is divided into 'Baseline' (not using ReShuffle-MS) and 'ReShuffle-MS' (use ReShuffle-MS).'model' provides models and performance reports for all classifiers. For the 'CNN' model, 'training_log.txt' is further provided.

本仓库包含论文《ReShuffle-MS》中所呈现的研究对应的数据集与模型。 'data.zip':内含上海第六人民医院的大肠杆菌(Escherichia coli)数据集(1)。其中raw文件夹下存储了格式为.txt文件的基质辅助激光解吸电离飞行时间质谱(MALDI-TOF mass spectra)数据,每个txt文件的文件名即为'SPEC_NUM'。每份txt文件的第一列为质荷比(mass-to-charge ratio),第二列为对应的信号强度(2)。'preprocessed'文件夹存储了经过预处理流程处理后的全部质谱数据,且已沿质荷比轴以3Da为步长完成分箱操作,最终得到6000个特征分箱。 本数据集采用分层抽样策略:将总数据的20%作为最终留出测试集(held-out test set),剩余80%的数据进一步划分为训练集(占比90%,即总数据的72%)与验证集(占比10%,即总数据的8%)(3)。文件'202403-11_E.coli.csv'展示了本批次大肠杆菌的抗生素敏感性测试(Antibiotic Susceptibility Testing, AST)结果。 'model.zip':内含本论文中使用的7种分类器模型,每种模型均分为“基线模型(Baseline,未使用ReShuffle-MS)”与“ReShuffle-MS模型(使用ReShuffle-MS)”两个版本。'model'文件夹提供了所有分类器的模型文件与性能报告。针对卷积神经网络(Convolutional Neural Network, CNN)模型,还额外提供了'training_log.txt'训练日志文件。

提供机构:
Zenodo
创建时间:
2025-09-17
二维码
社区交流群
二维码
科研交流群
商业服务