Consider this list of entries for the Tobacco Mosaic Virus (TMV). I selected to sort by 'Global pLDDT' from best to worst. It is my understanding that pLDDT is usually calculated per residue for a given prediction of a protein structure by comparing it to some 'true protein structure' (taken from PDB for example?).
I assume the 'Global pLDDT' is some form of aggregate of these per residue values? But what is the 'true protein structure' it is compared to? Other entries in the PDB?
pLDDT is a measure of confidence that ML models like AlphaFold provide for predictions of protein structures. The entries you are sorting here are not predictions, they are all experimental structures. As such, none of them have any pLDDT assigned. Notice that the order of your list does not change if you switch from 'Best to Worst' to 'Worst to Best' while sorting on Global pLDDT.
In the query builder for your link, notice the 'Include Computed Structure Models (CSM)' switch. It is set to not include predicted structures, although it appears the CSM providers RCSB integrates don't have any for TMV anyway. Here's RCSB's documentation for CSMs.
For experimental structures you probably want to sort by resolution, but the most important thing is to make sure the structure is exactly the molecule you think it is, e.g. not a mutant or a cleaved fragment, etc.
