Fig. 6.

Effect of the signal-to-noise ratio (SNR) on the precision-recall curves. Three synthetic datasets were generated through computational simulation of micrographs containing single-particle images with SNRs of 0.01, 0.008, 0.005, 0.003, 0.002 and 0.001. For each case, the CNN was first trained on the synthetic dataset of a given SNR and then used to examine the precision-recall relationship using another synthetic dataset with the same SNR. All synthetic datasets used the 70S ribosome as the single-particle model