Placeholder layout. Outward is decreasing identity to the seed, ranked;
sector is source organism; size is 1/resolution; colour is method; an amber halo
means ligand-bound. Phase 3 replaces the positions with a real structural
embedding from the pairwise TM-score matrix.
Entries
Construct coverage
How many of the family's constructs contain each residue of the seed
Coverage and disorder
Two different questions on one axis: what was in the construct,
and what was actually resolved
Rarely resolved regions
Present in the crystal, absent from the density
Seed sequence
Coloured by how often each residue is resolved
What people actually made
Domain architecture
Orthologues
Which organisms have structures, at what coverage and quality
Phase 3
Structures
See it, superpose it, and make the hero map mean what it says.
Mol* viewer, single entry and family superposition
Colour by entry or by conservation; AlphaFold overlay with pLDDT colouring
Pairwise TM-score matrix as a clustered heatmap with a dendrogram
The map's node positions move to a real structural embedding, and the clusters it finds (apo vs holo, open vs closed) become first-class filters
What phase 1 can already tell you
Phase 3
Ligands
Every chemical component in the family, and an honest verdict on which of them anyone meant to be there.
CCD ID, formula, SMILES, 2D depiction and occurrence count
Explicit ligand vs cryoprotectant vs buffer vs ion classification: PEG, glycerol, sulphate, Tris, MES, MPD and cacodylate flagged and separable
Cross-references to ChEMBL and DrugBank where they exist
What phase 1 can already tell you
Phase 3
Contacts
PLIP run family-wide, so binding-site contacts are comparable across entries rather than one-off.
Per-entry interaction diagrams
Family interaction fingerprint heatmap (ligand × residue)
A “hot residue” ranking, mapped back onto the alignment and onto Mol*
Runs as a background job with progressive fill: the map never waits for PLIP
Phase 3
Crystals
What actually worked, parsed out of _exptl_crystal_grow across the whole family.
Structured precipitant, salt, buffer, pH, temperature, method and additives
pH against precipitant class, coloured by resolution
A “what worked” summary table, exported in a shape the Top96 crystallisation predictor can ingest
What phase 1 can already tell you
Phase 3
Quality
A blunt traffic-light triage of which entries to trust.
Clashscore, RSRZ outliers, Ramachandran and rotamer outliers
R-free minus R-work gap
EDS, structure-factor and raw-data availability
What phase 1 can already tell you
Phase 4
Topology
Fold cartoons generated in-house, plus everything needed to hand the family to somebody else.
Topology diagrams from DSSP as custom SVG: strands as arrows, helices as cylinders, connectivity preserved (built in-house, not scraped from PDBsum)
Assembly and oligomeric state: author-assigned versus PISA-predicted, with disagreements flagged
Dossier export: one-click self-contained HTML and PDF family report
Shareable permalinks with the filter state encoded in the URL
Filing
Assembling the family from the RCSB, UniProt and InterPro.