MS2Query results 20-fold cross validation

The zip file contains the raw results of the 20-fold cross validation, the raw data downloaded from GNPS on 01-11-2022 and all the test data splits made. The raw results are stored in JSON format. The results of MS2Query and the benchmarking methods (cosine, modified cosine and MS2Deepscore) are stored. For each test spectrum three values are stored. The first value is the predicted score, the second value the tanimoto score between the correct annotation and the prediction and the last value is a boolean showing if the predicted spectrum was an exact 2D structure match. The figures in the MS2Query paper can be reproduced using the functions in https://github.com/iomega/ms2query/blob/main/ms2query/benchmarking/create_accuracy_vs_recall_plot.py该压缩包包含20折交叉验证的原始结果、2022年11月1日从GNPS（Global Natural Products Social Molecular Networking）下载的原始数据，以及所有已构建的测试数据集划分结果。原始结果以JSON格式存储。本数据集同时收录了MS2Query与各类基准测试方法（余弦相似度、改进型余弦相似度及MS2Deepscore）的运行结果。针对每条测试谱图，共存储三项数值：第一项为预测得分，第二项为正确注释结果与预测结果之间的塔尼莫特相似度得分，第三项为布尔值，用于标识预测谱图是否与真实结果存在精确二维结构匹配。可通过调用https://github.com/iomega/ms2query/blob/main/ms2query/benchmarking/create_accuracy_vs_recall_plot.py 中的代码函数，复现MS2Query论文中的相关图表。