Oncofusion

start / graph / protein-alk-domains

Proteinstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.king

ALK (anaplastic lymphoma kinase) domain architecture

also called ALK · anaplastic lymphoma kinase

ALK is the 3' partner of NPM1::ALK and the target of crizotinib, alectinib and lorlatinib. This entry lists where each of its functional parts sits along the protein, measured as residue numbers, so the fusion diagrams can draw which parts survive the break and which are thrown away.

The same thing, for a scientist

Reference domain architecture for ALK (UniProt Q9UM73, 1620 aa). A receptor tyrosine kinase with a large MAM/LDLa/MAM ectodomain and a 1116-1392 kinase domain. Created so the 3D fusion viewer can resolve fusion-npm1-alk, which previously had no coordinates for this partner and silently dropped the lane. Domain fate in the fusion is asserted by the fusion entry, not here.

What it looks like
ALK tyrosine kinase receptor1,620 aa
MAM 1MAM 1 · residues 264–427 · UniProt folded-domain call.LDL-receptor class A · residues 437–473 · UniProt folded-domain call.MAM 2MAM 2 · residues 478–636 · UniProt folded-domain call.Protein kinaseProtein kinase · residues 1116–1392 · UniProt folded-domain call.Heparin-binding region · residues 48–70 · UniProt functional region, not a fold boundary. Soft edges; render as low confidence.EGF-like · residues 987–1025 · UniProt functional region, not a fold boundary. Soft edges; render as low confidence.05001,0001,500
loading AF-Q9UM73-F1…
AlphaFold AF-Q9UM73-F1 ↗
colourshape

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

very high (90+)confident (70–90)low (50–70)very low (<50)
I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
Why this entry exists

The join checker reported fusion-npm1-alk: no protein entry for ALK. Without coordinates the comparative 3D viewer cannot draw the lane at all, and it was omitting it without warning rather than flagging it. This is a reference-coordinate entry: it asserts the architecture and nothing about the fusion.

Folded domains versus functional regions. UniProt mixes two very different annotation kinds. Domain, Zinc finger and DNA binding are fold calls with real boundaries. Region is a functional assertion that often spans disorder and can overlap other regions or run for hundreds of residues. Both are recorded with boundary_precision marking which is which, because drawing a soft functional region as a crisp box is exactly the kind of false precision this graph exists to avoid.

Architecture
PartResiduesKindPrecision
MAM 1264-427Domainfirm
LDL-receptor class A437-473Domainfirm
MAM 2478-636Domainfirm
Protein kinase1116-1392Domainfirm
Heparin-binding region48-70Regionsoft
EGF-like987-1025Regionsoft

79 PDB entries reference this protein.

Facts and figures
uniprotQ9UM73
gene symbolALK
length AA1620
protein nameALK tyrosine kinase receptor
alphafold IDAF-Q9UM73-F1
n PDB total79
uniprot verifiedtrue
accession resolutionexact gene-name search, reviewed human only, gene symbol confirmed in the response
created forfusion-npm1-alk
domains
Heparin-binding region48.0Heparin-binding region: 48MAM 1264.0MAM 1: 264LDL-receptor class A437.0LDL-receptor class A: 437MAM 2478.0MAM 2: 478EGF-like987.0EGF-like: 987Protein kinase1,116Protein kinase: 1116
domains — the numbers behind the chart
[
  {
    "name": "MAM 1",
    "start": 264,
    "end": 427,
    "feature_type": "Domain",
    "note": "UniProt folded-domain call.",
    "boundary_precision": "firm"
  },
  {
    "name": "LDL-receptor class A",
    "start": 437,
    "end": 473,
    "feature_type": "Domain",
    "note": "UniProt folded-domain call.",
    "boundary_precision": "firm"
  },
  {
    "name": "MAM 2",
    "start": 478,
    "end": 636,
    "feature_type": "Domain",
    "note": "UniProt folded-domain call.",
    "boundary_precision": "firm"
  },
  {
    "name": "Protein kinase",
    "start": 1116,
    "end": 1392,
    "feature_type": "Domain",
    "note": "UniProt folded-domain call.",
    "boundary_precision": "firm"
  },
  {
    "name": "Heparin-binding region",
    "start": 48,
    "end": 70,
    "feature_type": "Region",
    "note": "UniProt functional region, not a fold boundary. Soft edges; render as low confidence.",
    "boundary_precision": "soft"
  },
  {
    "name": "EGF-like",
    "start": 987,
    "end": 1025,
    "feature_type": "Region",
    "note": "UniProt functional region, not a fold boundary. Soft edges; render as low confidence.",
    "boundary_precision": "soft"
  }
]
PDB IDS
[
  "2KUP",
  "2KUQ",
  "2XB7",
  "2XBA",
  "2XP2",
  "2YFX",
  "2YHV",
  "2YJR",
  "2YJS",
  "2YS5",
  "2YT2",
  "3AOX",
  "3L9P",
  "3LCS",
  "3LCT",
  "4ANL",
  "4ANQ",
  "4ANS",
  "4CCB",
  "4CCU",
  "4CD0",
  "4CLI",
  "4CLJ",
  "4CMO"
]
How this could bite us back
  • Nothing here licenses targeting any part of ALK. These are coordinates, not a druggability claim; a domain with crisp boundaries is not thereby a ligandable pocket.
  • Drawing UniProt `Region` features as boxes in the viewer risks implying structural knowledge that does not exist. They are marked soft and should be rendered differently from fold calls.
Still open
  1. Where does the fusion-npm1-alk breakpoint fall relative to these boundaries, and is the junction codon actually mapped or only inferred from cDNA?
  2. Which of these parts are retained in the fusion, and is retention ever being confused with requirement for transformation?
  3. Do InterPro and the deposited crystal constructs agree with these UniProt boundaries? Where they do not, the disagreement should be recorded rather than silently resolved in UniProt's favour.
Every source behind this entry (1)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Human ALK (ALK tyrosine kinase receptor) is 1620 residues (UniProt Q9UM73) with the domain and region boundaries tabulated in this entry, and is referenced by 79 PDB entries.

    What this evidence cannot show

    Accession resolved by exact gene-name search restricted to reviewed human entries and confirmed against the gene symbol returned, because a previous lookup in this project silently resolved to the wrong gene. Coordinates are for the canonical isoform only; several of these proteins have isoforms that shift numbering. UniProt `Region` features are curator functional calls, not folds.

    Evidence that argues against this (1)
    • Domain boundaries are not a settled fact and differ between sources. InterPro, Pfam, SMART and individual crystal-structure constructs routinely disagree with UniProt by tens of residues, and published figures often use construct boundaries rather than database ones. Any downstream claim that a breakpoint falls just inside or just outside a domain is therefore only as firm as the boundary definition it assumes, which for every `Region` feature here is soft. InterPro Consortium 2026
Connected to