Fig. 3: Enzyme function prediction within the proteome of Prochlorococcus marinus MED4. | Communications Biology

Fig. 3: Enzyme function prediction within the proteome of Prochlorococcus marinus MED4.

From: Improved enzyme functional annotation prediction using contrastive learning with structural inference

Fig. 3

a Comparison of EC numbers annotations for P. marinus MED4 proteome through three methods: manual annotation from UniProt (bottom) and computational prediction using CLEAN-Contact (left) and CLEAN (right). The overlapping regions indicate EC numbers shared between different methods. b Assessment of prediction confidence between CLEAN-Contact and CLEAN for the n=583 annotated enzymes. Box plot elements: The center lines represent the median prediction confidences. The upper and lower box limits represent upper and lower quartiles, respectively. Whiskers extend to the maximum and minimum prediction confidences. Individual points represent all prediction confidence values. c Distribution of first-level EC numbers predicted by CLEAN-Contact for 1212 unannotated and unreviewed enzymes. d Visualization of 38 enzymes with high-confidence predictions and their corresponding predicted first-level EC number.

Back to article page