Ticket #17792: difference_summary.txt

The SKEMPI‐version of 1KBH has been “flat­tened” in a way that breaks the PDB specification. Here’s what we see when we compare the two files:

No header or metadata

The original 1kbh.pdb begins with HEADER, TITLE, COMPND, SOURCE, CRYST1, etc. SKEMPI’s SKEMPI_v2.0_1KBH.pdb starts immediately with ATOM lines and has zero HEADER, TITLE, COMPND, SOURCE or CRYST1 records.

All 20 NMR models concatenated into one

1KBH is an NMR structure with 20 models; in the official PDB file you’ll find 20 MODEL 
 ENDMDL blocks (40 lines total). In the SKEMPI file, those 20 blocks have been stripped of their MODEL/ENDMDL delimiters, so you get 33 060 ATOM records (20 × about 1 650 atoms) all in one continuous “model.”

Duplicate atom and residue numbers

Because each original model’s ATOM records (and their atom serial numbers) have simply been appended back-to-back, the file contains multiple atoms with the same serial numbers (1, 2, 
 1653, then 1, 2, 
 1653 again, etc.) and the same residue identifiers in each “block.” Most parsers will reject or confuse these duplicates.

Missing TERs and no END

The original has two TER records per model (40 TERs); SKEMPI leaves just two TERs (one per chain) at the end of the entire file. There is also no final END record.

Because of these violations of the fixed-column format and of the multi-model conventions (and because atom serials are no longer unique), many molecular viewers and parsers simply give up on SKEMPI_v2.0_1KBH.pdb.