1YCR is MDM2’s N-terminal domain (chain A, 109 residues) bound to a 15-residue transactivation-domain peptide from p53 (chain B). It was picked as the example because that shape – a short peptide docked onto a larger target protein – mirrors an AMP-vs-target-protein docking result, so the output here is a reasonable preview of what you’ll get running this tool on your own docked complexes.
Reference: Kussie et al. (1996) Science 274:948-953, doi:10.1126/science.274.5289.948. Coordinates downloaded directly from RCSB (files.rcsb.org/download/1YCR.pdb); PDB coordinate data is not copyrightable and is freely redistributable.
data/1YCR_MDM2_p53.pdb – the input structure (header trimmed of refinement-statistics
boilerplate to keep the file small; all ATOM/HELIX/SHEET/SEQRES records are intact).output/1YCR_MDM2_p53/ – everything the script produces from that input.python src/pdb_interface_analyzer.py examples/data/1YCR_MDM2_p53.pdb \
--ligand-chain B --receptor-chain A \
--ligand-label p53_peptide --receptor-label MDM2 \
--output examples/output
Complex_Interface_Report.txt lists 11 hydrogen bonds, 1 salt bridge, and 11
hydrophobic contacts. The hydrophobic-contact list correctly recovers the interface’s
three headline residues from the original paper – Phe19, Trp23, and Leu26 – as
the peptide’s main contacts burying into MDM2’s hydrophobic cleft.interface_summary_bubble.png / interface_contact_heatmap.png /
interface_interactions_network.png visualize the same contacts three different ways.secondary_structure.png shows the p53 peptide folding into a single alpha helix upon
binding (matching HELIX 5 in the source file) while MDM2 shows its four native helices.ramachandran_plot.png + Ramachandran_Summary.txt show backbone phi/psi angles;
~86-91% of residues fall in the illustrative favored regions here, with the outliers
concentrated at loop/turn regions – as expected for this resolution (2.6 A) structure.docked_complex_3d.html – open this directly in a browser for an interactive,
rotatable 3D view with toggleable chains, bonds, and render styles.Your own docking output will very likely differ in one respect: real docking results
usually have no HELIX/SHEET header records at all (that information comes from
crystallographic refinement, which a docking run doesn’t do), so secondary_structure.png
will show your chains as plain coil and say so on the plot. That’s expected, not an error.
data/antibody_epitope_BCR.pdb is a modeled antibody Fv (Heavy chain H, Light chain L)
bound to a 16-residue epitope peptide (chain A). It’s the multi-chain case: one side of
the interface is two chains acting as one unit, which is exactly what an antibody-antigen
or B-cell-receptor complex looks like. This file also has explicit hydrogen atoms, which
the analyzer automatically excludes from contact geometry (see the README’s “Important
caveats” section).
Interface analysis (both antibody chains treated as one receptor group):
python src/pdb_interface_analyzer.py examples/data/antibody_epitope_BCR.pdb \
--receptor-chain H,L --ligand-chain A \
--receptor-label Antibody --ligand-label Epitope \
--output examples/output
Single-chain analysis (chain H alone, to show the disulfide/Ramachandran/composition output):
python src/pdbsum_single_chain_analyzer.py examples/data/antibody_epitope_BCR.pdb \
--chain H --label "Antibody Heavy Chain" \
--output examples/output_single_chain
output/antibody_epitope_BCR/)interface_summary_bubble.png shows three bubbles – Epitope (A), Antibody (H),
and Antibody (L) – each sized by its own interface residue count, with separate
H↔A and L↔A connection lines so you can see the contact split between the two
antibody chains at a glance, instead of one lumped Antibody↔Epitope number.Complex_Interface_Report.txt finds 24 hydrogen bonds, 10 salt bridges, and
3 hydrophobic contacts, with a per-chain breakdown showing 9 of the 10 antibody
interface residues on chain H and only 1 on chain L – a realistic CDR-H3-dominant
paratope pattern, correctly attributed to the right chain even though H and L reuse
overlapping residue numbers (e.g. both have a residue 49).interface_interactions_network.png and interface_contact_heatmap.png both
show a dashed divider separating the chain-H block from the chain-L block, with residue
labels like Tyr49(L) – the (chain) suffix only appears when a side has more than
one chain, so the single-chain 1YCR example above doesn’t show it.docked_complex_3d.html – H-bond cylinders and interface highlighting are keyed to
the correct specific chain, not just a residue number, so nothing gets misattributed
between H and L.output_single_chain/antibody_heavy_chain/)Chain_Structure_Report.txt correctly detects both of an antibody Ig-domain’s
canonical disulfide bonds (Cys22-Cys96 and Cys144-Cys204, both “Formed”), reports
secondary structure as all-coil (this model has no HELIX/SHEET header records), and
finds ~89% of residues in the illustrative Ramachandran favored regions.secondary_structure_disulfides.svg – the PDBsum-style wiring diagram. Only the
Cys144-Cys204 pair is drawn as an arc: it and its partner land in the same wrapped row
of the diagram, while Cys22-Cys96 spans two different rows and so is listed in the
report only, not drawn (connecting an arc across a line wrap would need a much more
complex curve, and a misleading one felt worse than none).residue_composition.png breaks the 178-residue chain down by physicochemical class
(29% aliphatic, 30% polar, 14% Pro/Gly, etc).output_single_chain/ubiquitin/)Unlike the antibody example above, 1UBQ’s header has real HELIX/SHEET records, so
this is the example to look at for what the tool shows on a file with genuine secondary
structure:
Chain_Structure_Report.txt reports 21.1% helix, 43.4% sheet, 35.5% coil – matching
the known fold (3.5 turns of alpha helix plus a five-stranded mixed beta sheet) – and
0 disulfide bonds (correct: ubiquitin has no cysteines at all).secondary_structure_disulfides.svg shows the full topology in sequence order –
E1 (1-7), E2 (10-17), H1 (23-34), E3 (40-45), E4 (48-50), H2 (56-59), E5 (64-72) – with
purple sinusoidal helix ribbons and gold strand arrows, plus a handful of detected
beta/gamma turns in the connecting loops.secondary_structure_wiring.png shows the same topology as a simpler schematic
(shapes only, no sequence letters) – useful for a quicker at-a-glance view.web/antibody_epitope_visualizer.html is a zero-install, drag-and-drop version of the
interface analysis specifically for the H+L-vs-A case – open it directly in a browser,
no Python needed. See the main README for the scope difference in its
“non-bonded contacts” count.