The WHAT_CHECK output has been divided in a series of sections; each section dealing with a certain group of topics.
Sections 1 and 2 holds a series of checks that test for things that can influence many of the checks to follow in the following sections. If, for example, the symmetry cards are wrong then many checks that rely on a correct calculation of that symmetry will fail, or if WHAT_CHECK (actually the built-in PRODRUG) cannot interpret a ligand properly, then all kinds of ligand related checks will not work properly later on.
Section 3 holds a series on options that don't really check things, but that neveretheless can be very informative. This section holds, for example, Ramachandran plots. As has become clear from the early days of PROCHECK, the Ramachandran plot is one of the best protein quality indicators.
Other descriptive output is the secondary structure. In the PDB file 4NLL the secondary structure looks OK. Helices and strands have some length, and the percentages helix and strand seem OK. Obviously, unstructure proteins do exist, but their structure will (by definition!) not be solved...
Section 4 holds a series of coordinate administration checks. Unexpected atoms are listed, B-factors and occupancies are checked for plausibility, etcetera.
This section holds a series of nomenclature checks, like swapping Val Cγ1 and Cγ2, or the use of old-fashioned atom names, or weird chain identifiers.
Section 6 deals with geometric problems such as unusal bond lengths, bond angles, chiralities, tau angles, planarities, etcetera.
Section 7 deals with torsion angles, rotamers, backbone and side chain conformations.
Section 8 deals with bumps.
Section 9 deals with packing and inside/outside distribution considerations.
Section 10 deals with hydrogen bonds, waters, ions, flips of Asn, Gln, and His side chains.
The last two sections give a summary of the global statistics in terms of RMS-Z scores, quality scores, and normality scores, and the references