EGFR UniProt Sequence and PDB Structure Inspection Tool
Open human EGFR UniProtKB P00533 and inspect its sequence alongside PDB 4UV7, a crystal structure of the EGFR extracellular domain bound to a Fab fragment. The initial selection names residue 790 and the T790M substitution in UniProt numbering, which fall in the kinase domain and lie outside this ectodomain structure, so the page uses that gap to show how sequence, structural coverage and variant annotation are separate evidence layers. The workspace does not predict drug response from a highlighted residue.
The native protein workspace retrieves the cited UniProt entry and experimental structure, preserving source and retrieval metadata. PDB 4UV7 is an X-ray structure (2.1 Å) of the EGFR extracellular domain in complex with the Fab fragment GC1118A; SIFTS maps its EGFR chain to UniProt residues 25–645, so the kinase-domain residue 790 is not present. Selection 790 and substitution T790M use the reference sequence coordinate system, and the workspace shows them on the sequence while reporting that they lie outside this structure. A kinase-domain entry such as 3IKA is where residue 790 is resolved.
Establish the reference sequence before the substitution
A substitution string contains an original residue, a position and a replacement residue. It is meaningful only against a particular reference sequence. The first useful check is whether the stated original residue occurs at the named position. If it does not, investigate the accession, isoform and numbering convention instead of silently accepting the label.
EGFR is a useful example because the biological protein, processed forms and structural constructs can be described using different coordinate contexts. The initial T790M label uses UniProt P00533 coordinates. A mature-protein numbering convention must not be mixed with that reference. Preserve accession, sequence version and any isoform suffix with an exported annotation.
Treat structural coverage as a mapping question
The selected experimental structure is one record with its own construct, chains, resolved residues and experimental conditions. A sequence viewer may show the full reference while the three-dimensional model contains only a subset. Missing coordinates are not evidence that a residue is biologically absent; the construct may omit it or the experiment may not resolve it.
Inspect chain identity and mapping before measuring or interpreting a selected position. Author numbering, insertion codes and engineered changes can complicate a direct comparison. The structure record supplies evidence about what was measured, while the mapping connects that evidence to the reference sequence. A protein name match alone cannot replace those checks.
Keep annotation, structure and biological consequence distinct
A highlighted variant can help organise a review of nearby residues and available annotations. It does not establish a causal mechanism by its visual proximity to a ligand or another feature. Any functional claim needs evidence appropriate to the proposed mechanism, and a clinical claim requires evidence beyond a structural display.
The native workspace may retrieve external annotations with their own versions and coverage rules. Preserve those source labels and refusals. A missing score is not a neutral score, and a numerical predictor is not an experimental result. The page is designed to make these evidence layers inspectable while avoiding an unsupported diagnosis or treatment recommendation.
Equations and variables
Substitution reference check
reference[position] = stated original residue
- reference
- The selected UniProt accession and sequence version
- position
- One-based residue coordinate on that reference
- stated original residue
- The first amino-acid symbol in a substitution such as T790M
Mapped structural coverage
coverage = mapped reference residues / reference length
- mapped reference residues
- Reference positions with an explicit mapping to the selected structural context
- reference length
- Length of the stated sequence, not an arbitrary PDB chain
Scope and limitations
- UniProt precursor numbering, mature-protein numbering and author PDB residue labels may differ. Do not subtract a signal-peptide length or apply an offset without an explicit mapping.
- An experimental structure can cover a domain or engineered construct and may contain mutations, missing residues, ligands or chain differences. It is not automatically a full-length wild-type receptor model.
- A variant annotation or model score does not by itself establish pathogenicity, inhibitor sensitivity, resistance or a clinical treatment decision. Missing annotations remain missing.
Sources and provenance
- UniProtKB human EGFR P00533CC BY 4.0
The interactive workspace displays the release of the retrieved record.
- RCSB PDB experimental entry 4UV7: EGFR extracellular domain with the Fab GC1118ACC0; existing wwPDB/PDB source classification
Entry title and SIFTS chain mapping (UniProt 25–645) read on 17 October 2026.
- PDBe SIFTS sequence–structure mappingExisting PDBe/SIFTS source terms; mapping reference
Technical questions
Why might residue 790 not have the same label in a PDB chain?
UniProt and a PDB author can use different coordinate conventions, and an experimental construct may cover only part of the protein. Use the explicit mapping and chain identity. A convenient arithmetic offset is not sufficient evidence that two residue labels refer to the same position.
Does selecting T790M modify the experimental structure?
No. A substitution annotation identifies a reference-sequence change for inspection; the workspace does not remodel side chains. Residue 790 is in the kinase domain and is not present in the 4UV7 ectodomain structure at all, so the workspace reports it as outside coverage rather than drawing it. A kinase-domain entry is needed to see the residue in 3D. Read the selected PDB entry before interpreting the geometry.
Can the displayed variant information predict inhibitor response?
Not on its own. Sequence identity, structural context and any external score address narrower questions than clinical response. Drug-specific biochemical and clinical evidence is needed for such a conclusion, and the workspace does not invent that evidence from a highlighted residue.
Connected workflows
- KRAS G12D trial-record exploration
Connect target identity to dated clinical-study records.
- Protein A280 concentration calculation
Use the exact construct sequence for protein amount calculations.
- CNS property inspection and MPO limitations
Keep structural and descriptor evidence separate from biological predictions.
- AlphaFold pLDDT and PAE Domain Confidence Inspector
Read the predicted structure confidence metrics (pLDDT and PAE) for the same protein.
- UniProt to PDB SIFTS Residue Numbering and Chain Mapper
Resolve UniProt-to-PDB residue numbering through SIFTS before reading a position.
- Kinase Active Site and Ligand Binding Pocket Inspector
Examine the kinase ATP pocket and gatekeeper residue in 3D.