Show pseudobonds between all adjacent residues colored by AlphaFold PAE score

[Tom Goddard]

To see interfaces where AlphaFold has low confidence it could be useful to show a pseudobond between every pair of contacting residues joining the CA positions and colored according to the PAE plot score. Might want to filter to just show the high error pseudobonds.

[Tom Goddard]

Might be even better to just take every pair of nearby residues within a selection, or between selected and not selected.

[Tom Goddard]

I added a command to make the colored pseudobonds

alphafold contacts /A [toResidues /B] [distance 3] [flip false] [palette paecontacts] [range 0,20] [radius 0.2] [dashes 1]

The residues must belong to an alphafold structure which has PAE data already opened. It finds residues having any atoms within the specified distance (default 3 Angstroms). It uses the paecontacts colormap by default which is like the "pae" colormap used for heatmaps only it ends in red instead of in white.

0,blue:5,cornflowerblue:10,yellow:15,orange:20:red

The PAE matrix is not symmetrical. The first command argument specifies the aligned residue and the toResidues argument specifies the residue whose error is reported. The "flip true" option swaps the two sets of residues (equivalent to swapping the initial residues argument and the toResidues argument). If toResidues is not specified then it defaults to all the structure residues not in the the first argument residue spec. In other words it finds all residues contacting the first argument. If the first argument specifies all residues (e.g. #1) then the second argument defaults to all residues, showing pseudobond for all close residue pairs in the structure (bidirectionally).

It can be useful to color using a smaller range of PAE values to highlight differences among high confidence contacts, for example "range 1,5".

The radius (default 0.2) and number of dashes (default 1 = solid pseudobond) for the pseudobonds can be specified.

[Tom Goddard]

Here's an example

alphafold fetch tacan_human
alphafold pae #1 uniprot tacan_human
alphafold contacts #1 distance 8

I've attached an image. It shows that two long helices that seem to be detached from the main structure are in fact fairly confidently oriented, and in fact an experimental structure of the dimer shows those helices are in the right position.

[Tom Goddard]

Example tacan_human contacts image.