Predictions results of DeepViral

Predictions results of the DeepViral paper, available at https://doi.org/10.1101/2020.04.22.055095. All files are tab separated, with 6 fields: &lt;virus protein&gt; &lt;human protein&gt; &lt;virus taxon&gt; &lt;virus family&gt; &lt;predicted score&gt; &lt;label based on HPIDB&gt; 1. The prediction results for 14 virus families based on the LOFO experiments. Each method was evaluated 5 times, i.e., indexed 0-4. The baseline methods: RCNN - preds_rcnn_i.txt and RF+Doc2Vec - preds_rf_i.txt The DeepViral variants: pheno: seq + viral embedding go: seq + human embedding seq and joint as described 2. The prediction results for 4 virus species based on the LOSO experiments. Each method was evaluated 5 times, i.e., indexed 0-4. t - test virus; v - validation virus. Taxon IDs are based on NCBI Taxonomy. species - LOSO; family - LOFO本数据集为DeepViral论文的预测结果，论文可通过https://doi.org/10.1101/2020.04.22.055095获取。所有文件均采用制表符分隔格式，包含6个字段：<病毒蛋白> <人类蛋白> <病毒分类单元> <病毒科> <预测得分> <基于HPIDB的标签>。 1. 基于留一科淘汰（Leave-One-Family-Out, LOFO）实验得到的14个病毒科的预测结果。每种方法均进行了5次评估，索引编号为0至4。基准方法包括：RCNN——对应文件preds_rcnn_i.txt；RF+Doc2Vec——对应文件preds_rf_i.txt。DeepViral变体模型包括：pheno：序列+病毒嵌入；go：序列+人类嵌入；seq及联合嵌入（详见论文描述）。 2. 基于留一种淘汰（Leave-One-Species-Out, LOSO）实验得到的4个病毒物种的预测结果。每种方法均进行了5次评估，索引编号为0至4。其中t代表测试病毒，v代表验证病毒。分类单元ID基于NCBI分类学数据库。species字段对应留一种淘汰（LOSO）实验，family字段对应留一科淘汰（LOFO）实验。