Extended Data Fig. 3: The predicted S. aureus DltBDX structure is a high-confidence prediction.

a, (top) The top-ranked AlphaFold2 model of the S. aureus DltBDX complex, with proteins colored as in Fig. 2c, is a high-confidence model. The predicted local Distance Difference Test (plDDT) is a metric of model confidence, with plDDT values > 90 representing very high confidence, 70–90 representing moderate confidence, 50–70 representing low confidence, and < 50 representing very low confidence. The graph in the upper right shows the plDDT values at each residue for all five models. The numbering on the x-axis here corresponds to the residue number for the concatenated sequence of the complex (that is, the sequence of DltX concatenated N-terminally to the sequence of DltB, and then that overall sequence concatenated N-terminally to the sequence of DltD). The rightmost of the two models underneath the ‘Rank 1’ heading shows a graphical depiction of the plDDT values with a color gradient scale (dark blue representing a plDDT value of 100, and red representing a plDDT value of 0). (bottom) Each of the other four models is highly similar to the top-ranked model. Five models were generated by ColabFold, and the models were ranked by predicted template modeling (pTM) scores. b, The relative positions between the three proteins are high-confidence based on the Predicted Aligned Error (PAE). Here, the value at each (x, y) point in the graphs indicates the ‘expected position error at residue x if the predicted and true structures were aligned on residue y’ (https://alphafold.ebi.ac.uk/faq). Low values (blue) indicate low expected position error. Again, the numbering on the x-axis corresponds to the residue number for the concatenated sequence of the complex. Numbering on the y-axis (not shown) would place ‘0’ at the top of the axis. c, Views of the surface representation of the S. aureus DltBDX AlphaFold2 model.