Skip to main content

Table 1 The number of miRNA:target pairs in each dataset used in training, validating and testing the proposed approach

From: miTAR: a hybrid deep learning-based approach for predicting miRNA targets

Datasets Positive pairs Negative pairs Training set Validation set Test set
DeepMirTar dataset 3908 3850 4964 1241 1552
miRAW dataset 31,660 30,993 40,096 10,025 12,531
MirTarRAW dataseta 13,860 13,860 17,740 4435 5544
DeepMirTarIn datasetb 48 NA    
miRAWIn datasetb 929 890    
DeepMirTarLeft datasetc 443 385    
miRAWLeft datasetc 21,265 20,598    
  1. Blank box means not applicable. NA represents the data do not exist
  2. aDataset generated by combining the DeepMirTar and the miRAW dataset. Details are in “Methods” section
  3. bIndependent test dataset for evaluating different methods/models
  4. cThe surplus dataset from DeepMirTar and miRAW combining the two datasets. Details are in “Methods” section