Figure 4
From: CRYSTALP2: sequence-based protein crystallization propensity prediction
Prediction quality (y-axis) measured with accuracy and MCC (shown using markers) when training the CRYSTALP2 model on subsets of the FEAT training dataset (x-axis) and testing on the TEST-NEW, TEST and TEST-RL datasets. The solid and dashed lines show linear regression trends associated with the increasing size of the training dataset for the accuracy and MCC, respectively.