patGPCR: A Multitemplate Approach for Improving 3D Structure Prediction of Transmembrane Helices of G-Protein-Coupled Receptors

Figure 5

Statistical comparison between patGPCR after a 50-fold bootstrap estimation and the SWISS-MODEL on FL, TM, and Loop RMSD. 𝑥-axis indicates 1-percentile average RMSD of 50-fold bootstrap estimation of patGPCR prediction results and 𝑦-axis indicates the RMSD of the SWISS-MODEL. The error bars of the points indicate standard deviation 1 percentiles of patGPCR. Red, green, and blue show the comparison on TM, FL, and loop RMSD values, respectively. Points below the line represent targets where patGPCR yields higher accuracy, and above the line the SWISS-MODEL yields higher accuracy.