Skip to main content

Table 4 Performance of SWnet in terms of MSE in random split dataset by cell lines with different r-radius subgraphs

From: SWnet: a deep learning model for drug response prediction from cancer genomic signatures and compound chemical structures

Weight layer Self-attention r-radius MSE \(R^2\)
Single No 1 2.3090 0.6756
Single No 2 2.3102 0.6755
Single No 3 2.2793 0.6799
Multi Yes 1 2.2975 0.6773
Multi Yes 2 2.2840 0.6792
Multi Yes 3 2.3304 0.6726
  1. Bold values represents the result is the best performance among the models participating in the comparison