Figure 3.

The impact of local model quality on MR success ratio in the function of global model quality. The success ratio is the fraction of correct MR solutions found by models of a given type. Among the model types are: 1) models with B-factor vales set to 20: ALL_20, BACKBONE_20. 2) models modified according to real accuracy of a model: ALL_IDEAL, BACKBONE_IDEAL, ALL_CA_IDEAL, BACKBONE_CA_IDEAL. 3) models modified according to predicted local quality by MQAPs: MetaMQAP-evaluated, MetaMQAPclust-evaluated; 4) templates converted into polyalanine models. The black, grey and blue bars correspond to search models of high (GDT_TS≥80), moderate (70≤GDT_TS<80) and low (GDT_TS<70) global quality.