SIT1 - one of only three surface proteins in the screen that is both expressed in KMT2A::AFF1 lines and absent from normal parenchyma, and it is clean because it is leukocyte-restricted, not because it is leukaemic
also called SIT · SHP2-interacting transmembrane adapter protein · gp30/40
SIT1 is a small protein on the surface of white blood cells. It is one of only three surface proteins in this whole screen that is present in the leukaemia and essentially absent from normal organs - but it is absent from organs simply because it belongs to white blood cells, which is also why it cannot be specific to this leukaemia.
The same thing, for a scientist
SIT1 (UniProt **Q9Y3P8**, SIT1_HUMAN, **196 aa**, ENSG00000137078) is a transmembrane adapter with an extracellular topological domain, annotated Cell membrane [E1]. In the screen it is one of **three** genes (with CD19 and S1PR4) that clear the expression filter in the KMT2A::AFF1 B-lineage arm and are parenchymally clean: B-arm mean **57.5 nTPM** (SEM 49.6, RS4;11 65.4), parenchymal maximum **2.7 nTPM in duodenum** [E2]. **And it is not fusion-specific.** 5.28x a B-precursor comparator but only **1.18x** the KMT2A-rearranged-other-partner arm, failing the pre-registered 2x partner threshold. Its normal-tissue profile correlates with PTPRC at **r = 0.882** - the highest of the three - which is the direct statement that its parenchymal cleanliness is a consequence of being a leukocyte protein [E2]. GTEx agrees: peak tissue **Cells_EBV-transformed_lymphocytes at 104.84 TPM**, whole-body median **0.33 TPM** [E3]. In normal lymphoid tissue it is high - thymus 87.5, lymph node 46.8, tonsil 30.7 nTPM - so targeting it would deplete T cells. **Recorded as a negative candidate**, and as the clearest worked example of why the three clean genes cannot also be fusion-specific: cleanliness in this screen is produced by leukocyte restriction, and leukocyte restriction is what a lineage antigen is.
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.
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.
An unregistered symbol collision, found and reported
gene_exact:SIT1 against UniProt returns two reviewed human records, and the first is not SIT1. Q9NP91 / SLC6A20, the sodium- and chloride-dependent transporter XTRP3, carries SIT1 as an alias - sodium/imino-acid transporter 1 - and is returned ahead of Q9Y3P8 4.
Resolved through HGNC, which assigns one primary record per symbol: SIT1 -> ENSG00000137078 / Q9Y3P8 (signaling threshold regulating transmembrane adaptor 1), and SLC6A20 -> ENSG00000163817 / Q9NP91 with SIT1 among its aliases 4. This beat's data keys on ENSG00000137078 and is correct.
This token has no row in data/symbol-collisions.json - checked with word-boundary matching before the first query, as the standing rule requires, and it returned nothing. It should be added, and this entry is the report. It is the same shape as the BACH1 row: a symbol that is a current primary symbol for one gene and an alias for another, where the obvious query returns the other one first.
| uniprot | Q9Y3P8 |
|---|---|
| length AA | 196 |
| ensembl | ENSG00000137078 |
| alphafold ID | AF-Q9Y3P8-F1 |
| hpa protein class | Predicted membrane proteins |
| b arm mean nTPM | 57.5 |
| b precursor comparator nTPM | 10.89 |
| ratio over b precursor | 5.28 |
| ratio over other kmt2a partner | 1.18 |
| parenchymal max nTPM | 2.7 |
| parenchymal peak tissue | duodenum |
| lymphoid profile r vs PTPRC | 0.882 |
| gtex v8 median of 54 tissues | 0.33 |
| screen verdict | F1 surface PASS, F2b expressed PASS, F4b parenchymal CLEAN, F3b fusion-enriched FAIL (1.18x other KMT2A partners) |
| symbol collision | SIT1 is also an alias of SLC6A20 (Q9NP91, 592 aa). NOT in data/symbol-collisions.json as of 2026-09-23; reported for addition. |
| shown for fusion | NO - nobody has looked in any fusion. SIT1 appears in this project only as an output of this screen; PubMed returns 1 record for SIT1[tiab] AND leukemia[tiab]. |
index arm nTPM — the numbers behind the chart
{
"SEM": 49.6,
"RS4;11": 65.4,
"MV4-11": 0.2
}normal lymphoid nTPM — the numbers behind the chart
{
"thymus": 87.5,
"lymph node": 46.8,
"tonsil": 30.7,
"spleen": 21.7,
"appendix": 17.8,
"bone marrow": 1.7
}gtex v8 peak
{
"tissue": "Cells_EBV-transformed_lymphocytes",
"tpm": 104.84
}- SIT1's clean normal-tissue profile is leukocyte restriction restated. Targeting it would deplete T cells at thymus 87.5 nTPM, which for a T-cell therapy is self-defeating and for any therapy is worse than B-cell aplasia.
- Listing SIT1 as one of three 'clean' hits invites it to be read as a candidate. It is recorded as a negative candidate, and the screen's own verdict on it is FAIL.
- Is SIT1 expressed on primary KMT2A::AFF1 blasts at all? Its entire basis here is two cell lines and PubMed returns one record intersecting SIT1 with leukaemia.
- Does the 196-residue protein present enough extracellular surface for an antibody? Short single-pass adapters are poorly represented among successful CAR targets and nobody has asked why.
- Should data/symbol-collisions.json gain a SIT1/SLC6A20 row? This beat reports it; the registry is not this beat's to edit.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
SIT1 is UniProt Q9Y3P8 (SIT1_HUMAN), signaling threshold-regulating transmembrane adapter 1, 196 amino acids, annotated Cell membrane with an extracellular topological domain.
What this evidence cannot show
UniProt release 2026_03. The topological domain annotation is inferred rather than experimental.
Evidence that argues against this (1)
- A 196-residue protein with a short extracellular domain may present too little epitope for an antibody or CAR to engage productively, independently of its expression. UniProt 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
SIT1 is 49.6 nTPM in SEM and 65.4 in RS4;11 (B-arm mean 57.5) with a parenchymal maximum of 2.7 nTPM in duodenum; it is 5.28x a 13-line B-precursor comparator but only 1.18x the KMT2A-rearranged-other-partner arm; its 51-tissue profile correlates with PTPRC at r = 0.882; normal thymus 87.5, lymph node 46.8, tonsil 30.7 nTPM.
analyses/kmt2a-aff1-surface-target-screen/ - PREREGISTRATION.md, run.py and results.json, this repository. Pre-registered before any expression value was inspected. · Oncofusion Knowledge Graph 2026What this evidence cannot show
Two index lines, one value per line, no replicates. It is 0.2 nTPM in MV4-11, so it fails the all-three-lines version of the expression filter entirely.
Evidence that argues against this (1)
- High expression in normal thymus, lymph node and tonsil means SIT1-directed therapy would deplete T cells - which for a cell therapy is self-defeating, and for any modality is a more serious immunosuppression than B-cell aplasia. Human Protein Atlas 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
GTEx v8 puts SIT1 at 104.84 TPM in EBV-transformed lymphocytes with a whole-body median of 0.33 TPM across 54 tissues.
GTEx v8 median gene expression, SIT1 (ENSG00000137078.8) across 54 tissues: Cells_EBV-transformed_lymphocytes 104.84 TPM, whole-body median 0.33 TPM. Fetched 2026-09-23 · GTEx Portal 2026What this evidence cannot show
The peak GTEx tissue is a cell line rather than an organ, which this project has recorded before as a recurring feature of leukocyte-restricted genes in GTEx.
Evidence that argues against this (1)
- A whole-body median of 0.33 TPM looks outstanding and is entirely a restatement of leukocyte restriction, not evidence of leukaemia specificity. GTEx Portal 2026
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
A UniProt gene_exact query for SIT1 returns SLC6A20 (Q9NP91) ahead of SIT1 (Q9Y3P8), because SLC6A20 carries SIT1 as an alias; HGNC assigns SIT1 primary to ENSG00000137078/Q9Y3P8 and SLC6A20 to ENSG00000163817/Q9NP91.
What this evidence cannot show
Observed on one date against one query form. Ordering of UniProt search results is not contractually stable, so the specific ordering may differ; the alias relationship is the durable fact.
Evidence that argues against this (1)
- A length check would catch this collision - 196 aa against 592 aa - so it is less dangerous than collisions between similarly sized proteins, which this project has recorded as the harder class. HGNC 2026
one of three parenchymally clean genes, clean because leukocyte-restricted (r vs PTPRC 0.882)
- expressed in →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
49.6 nTPM in SEM and 65.4 in RS4;11, but 1.18x other KMT2A partners - not fusion-specific