Fig. 8
From: Freeprotmap: waiting-free prediction method for protein distance map
Comparison between proposed R-former and other models. a, Experimental result of compared models on the 4.05_release dataset. b, Experimental result of compared models on CASP_14D dataset. c, Experimental result of compared models on CASP_14F dataset. d, Experimental result of compared models on CASP_15D dataset. Note: F(U-Net), F(UCTrans) and F(LR-L) denote the proposed method, FreeProtMap, implemented by using U-Net, UCTrans and L-layer CNN as the skeleton network instead of the R-former, respectively. F(R-former) denote the proposed method, FreeProtMap, implemented by using R-former as the skeleton network. LR-L denotes L-layer CNNs with the kernel sizes of \(1 \times 1\). The units of \(MAE_{d}\) and RMSE are all Å