Figure 4.

FRET X fingerprinting simulation results assuming optimal and suboptimal experimental conditions

FRET X fingerprint classifier cross-validation performance measures are shown for three combinations of tagged residue types, C, C + K, and C + K + R, and two labeling qualities, “optimal,” where all targeted residues and no off-target residues were labeled, and “suboptimal,” where erroneous labeling occurred following the rules in Table S3.

(A) Venn diagram showing numbers of proteins that were found to be well identifiable, i.e., that were correctly identified in more than 5 of 10 cross-validation folds. The total number of proteins is 311.

(B) The identification accuracy of proteins under optimal and suboptimal labeling conditions.

(C) Average classifier accuracy as a function of the number of tagged residues in structures, aggregated in five groups with similar numbers of tags. Whiskers denote ± one standard deviation. Accuracies for different resolutions and fsuboptimal labeling scenarios are shown in Figure S6.