Hi !

I am a PhD student at the University of Montpellier.

I am conducting a study on bacterial resistance to rifampin. Rifampin is an antibiotic that binds to a subunit of RNA polymerase. Resistant bacterial strains have been identified, and I am investigating whether the rifampin binding site is altered in the subunit of these bacteria. To address this, I modeled the 3D structure of a sensitive strain and a resistant strain using Colab AlphaFold2. I then used the PrankWeb 3 online tool to identify possible ligand binding sites in both cases, and I found the rifampin binding site in both the resistant and sensitive strains.

The PrankWeb 3 results show that in the sensitive strain's subunit, 18 amino acids are predicted by PrankWeb 3 in my site of interest, while only 14 amino acids are predicted in the resistant strain. The issue is that the 4 amino acids difference are exactly the same between the resistant and sensitive strains (no residue difference). Moreover, the subunits of the resistant and sensitive strains have very similar overall 3D conformations and sizes (with only a few amino acids substituted).

Therefore, I do not understand why the algorithm behind PrankWeb 3 identifies a difference in the binding site.

We hypothesize that microvariations in the protein sequence may influence electrochemical interactions, which could increase or decrease the number of amino acids in the binding site explaining difference between sensitive and resistant (but we are really not sure).

My question is: Is PrankWeb 3 reliable enough to identify residue-level differences even in this context, or can it produce inaccuracies regarding the residues?

Thank you for your time and assistance