No published biology exists for AFF1-AS1 under any name - and the 2026 review of non-coding RNAs in KMT2A-rearranged ALL does not mention it
also called Q155 AFF1-AS1 literature bound · antisense transcription at the AFF1 locus
The antisense RNA sitting 151 letters from AFF1's start has never been written about. Not in a paper title, not in an abstract, not inside the full text of any open-access paper, and not in the review published this year whose whole subject is non-coding RNAs in exactly this leukaemia.
The same thing, for a scientist
A bounded absence, with the queries, their translations and their denominators. **PubMed**, through `scripts/pubmed_safe.py` so no zero can be a silent instrument failure: `"AFF1-AS1"[tiab]` **0**, `AFF1-AS1[All Fields]` **0**, `"AFF1-AS"[tiab]` **0**, `"AFF1 antisense"[tiab]` **0**, `"AFF1 antisense RNA 1"[All Fields]` **0**, `"AF4 antisense"[All Fields]` **0**, `"MLLT2 antisense"[All Fields]` **0**; `LOC100506746[All Fields]` **1**, an ALS GWAS where the symbol is a locus label [E1]. **Europe PMC**, which indexes open-access **full text**: `"AFF1-AS1"` **hitCount 0** [E1]. **GEO** `db=gds`: `"AFF1-AS1"` returns `quotedphrasesnotfound: ["AFF1-AS1"]` - the phrase is not in the index at all, which is a stronger negative than a bare zero [E1]. **The instrument works.** Through the identical shapes: `AFF1[tiab]` 226, `AF4[tiab]` 1,219, `MLLT2[tiab]` 6, `"HOXA-AS2"[tiab]` 99, `HOTAIR[tiab] AND antisense[tiab]` 633, `KMT2A[tiab] AND lncRNA[tiab]` 9, `AFF1[tiab] AND leukemia[tiab]` 121; in Europe PMC `"HOXA-AS2"` 651 and `"AFF1"` 2,129; in GEO `"HOXA-AS2"` 24. **Zero instrument failures in 32 queries.** **And the reagent test, because a PubMed zero is bounded by abstracts.** The paper most likely to contain this transcript as a supporting item is **PMID 41358901**, *"The hidden regulators: Non-coding RNAs in KMT2A-rearranged acute lymphoblastic leukemia"* (Int J Cancer, 2026). Its full text was read from Europe PMC (PMC12922646, HTTP 200, `<body>` present, 56,262 characters after stripping tags first and unescaping second, **with every unicode dash normalised to ASCII before searching** - the document uses U+2010 and an ASCII-hyphen search would have returned a meaningless zero). It names **AFF1 25 times**, **AFF1-AS1 zero times**, **C4orf36 zero times**, and the lncRNA symbols it does discuss are HOXA10-AS, LAMP5-AS1, LINC00226, LINC00511, LINC00958, LINC01221 and LINC01226 - **seven, and not this one** [E2]. **The class of absence is `measured-not-analysed`, not `never-measured`**, and the distinction is worth three orders of magnitude here. Every RNA-seq library ever sequenced from a KMT2A::AFF1 sample contains reads over `chr4:86,924,893-86,934,850`. Nobody has looked at them.
What this absence does and does not bound
It bounds: that no paper indexed by PubMed or Europe PMC names this transcript under any of seven constructed names, and that the field's current review of exactly this topic does not know it exists.
It does not bound:
- a paper that describes the transcript by coordinate and never names it. Rule 70's case: there is no HGNC for an unnamed antisense RNA in a figure panel, and the names tried are listed above precisely so a reader can judge what the bound is worth.
- a deposited dataset in which it is quantified. GEO's keyword index does not reach inside a supplementary expression matrix. A stranded RNA-seq deposit from any KMT2A-r cohort almost certainly contains a row for it, unquantified by anyone.
- anything about whether the RNA is made. An annotation and a literature silence are compatible with a real, abundant, functional transcript and with nothing at all.
And the honest comparison: "IRX2-DT"[tiab] also returns 0 in PubMed and only 4 in Europe PMC, and Q153 wrote an entry about IRX2-DT anyway. Being unstudied is the normal state of a divergent antisense RNA; the reason AFF1-AS1 is worth an entry is not that the silence is unusual but that the gene it silences next to is the index fusion's own partner.
The general question: is antisense transcription at the AFF1 locus regulated?
Asked separately from the symbol, because rule 71's lesson is that a symbol search can be accurate and false. Every query below ran through the same tool with its translation recorded:
| query | count | what the hits are |
|---|---|---|
AFF1[tiab] AND antisense[tiab] | 1 | PMID 16703585, panhandle PCR breakpoint cloning - antisense is a primer-design word |
AFF1[tiab] AND lncRNA[tiab] | 1 | PMID 36595475, a leukaemia transcriptome resource - AFF1 as a fusion partner |
AFF1[tiab] AND "long noncoding"[tiab] | 1 | same record |
AF4[tiab] AND antisense[tiab] | 1 | PMID 31692047, ZFPM2-AS1 / miR-511-3p / AFF4 in lung adenocarcinoma - not AFF1 |
AFF1[tiab] AND divergent[tiab] | 1 | PMID 21436398, Conserved eukaryotic fusogens - a fusogen/AFF1 collision, and AFF1 there is the Caenorhabditis fusion protein, not the human gene |
AFF1[tiab] AND bidirectional[tiab] | 0 | |
AFF1[tiab] AND "antisense transcription"[tiab] | 0 | |
MLLT2[tiab] AND antisense[tiab] | 0 | |
AFF1[tiab] AND autoregulation[tiab] | 0 | |
AFF1[tiab] AND autoregulatory[tiab] | 0 | |
AFF1[tiab] AND promoter[tiab] | 17 | the only non-trivial set; none is about antisense or divergent transcription at the AFF1 locus |
Note the collision worth logging: AFF1 is also the name of a C. elegans fusogen (Anchor-cell Fusion Failure 1), and PMID 21436398 in Science is about that protein. A count of AFF1[tiab] is therefore contaminated at an unmeasured rate, and any absence built on an AFF1-keyed query should title-check. This is a new row for data/symbol-collisions.json and is reported rather than written, because that file is not this beat's to edit.
So: nobody has asked whether antisense transcription at the AFF1 locus is regulated or functional, in any system, under any of the vocabularies searched.
| absence class | measured-not-analysed |
|---|---|
| absence class justification | Every RNA-seq library from a KMT2A::AFF1 sample contains reads over chr4:86,924,893-86,934,850. The transcript is annotated, the data is public, and the analysis has never been run. Closing this costs compute, not wet work. |
| reagent test | PMID 41358901 / PMC12922646 full text read and searched with unicode dashes normalised; 0 mentions |
| n queries | 32 |
| n instrument failures | 0 |
| new symbol collision to log | AFF1 is also Anchor-cell Fusion Failure 1, a C. elegans fusogen (PMID 21436398, Science 2011). AFF1[tiab] counts are contaminated at an unmeasured rate. Reported for data/symbol-collisions.json; not written by this beat, which does not own that file. |
names searched
[ "AFF1-AS1", "AFF1-AS", "AFF1 antisense", "AFF1 antisense RNA 1", "LOC100506746", "AF4 antisense", "MLLT2 antisense" ]
indexes searched
[ "PubMed (esearch via scripts/pubmed_safe.py)", "Europe PMC REST search (abstract + open-access full text)", "NCBI GEO DataSets (db=gds)" ]
measurements — the numbers behind the chart
[
{
"quantity": "n_pubmed_records",
"subject": "AFF1-AS1 across seven constructed names",
"value": 0,
"units": "records",
"n": 7,
"universe": "PubMed esearch 2026-09-24 via pubmed_safe; LOC100506746 returns 1 off-topic record, all other names 0",
"null": 99,
"note": "null is \"HOXA-AS2\"[tiab] = 99, a comparable haematopoietic antisense RNA, through the identical shape. Rule of three on 0/7 names is not meaningful; the bound is the name list, not a sample size"
},
{
"quantity": "n_europepmc_fulltext_records",
"subject": "the quoted phrase \"AFF1-AS1\" in Europe PMC's abstract + open-access full-text index",
"value": 0,
"units": "records",
"n": 1,
"universe": "Europe PMC REST search 2026-09-24; controls \"HOXA-AS2\" 651, \"IRX2-DT\" 4, \"AFF1\" 2129",
"null": 651,
"note": "the tightest bound available: a full-text zero, not an abstract zero"
},
{
"quantity": "n_mentions_in_the_topic_review",
"subject": "AFF1-AS1 in the full text of the 2026 review of ncRNAs in KMT2A-rearranged ALL",
"value": 0,
"units": "mentions",
"n": 56262,
"universe": "PMC12922646, 56,262 characters after tag-stripping, unicode dashes normalised; AFF1 appears 25 times and seven other lncRNA symbols are named",
"null": 7,
"note": "null is the number of lncRNA symbols the review DOES name, i.e. the size of the field's current attention list"
}
]- A silence this complete is a reason to be careful, not a reason to be dismissive. An uncharacterised antisense RNA at the index fusion's own partner is exactly the object a 2 kb promoter readout would hide, and every promoter-scale intervention at AFF1 will hit it with no literature to predict the consequence.
- Equally, the silence means no reviewer or collaborator will have heard of it, so any claim this project makes about AFF1-AS1 stands entirely on its own annotation evidence and must not be written as though a field existed behind it.
- Is AFF1-AS1 detectable in stranded RNA-seq from a KMT2A::AFF1 sample? This is the single cheapest thing left in this beat: one strand-aware coverage query over chr4:86,924,893-86,934,850 in an existing deposit, no wet work, and it converts a measured-not-analysed absence into a measurement.
- Does the transcript respond to menin inhibition? If it is a fusion-driven divergent transcript it should track the fusion's departure, and GSE252378's treated arm plus any menin-inhibitor RNA-seq deposit would show it.
- How many of this graph's 72 absence entries would change class if the archive query were run? This one moved from looking like never-measured to being measured-not-analysed on a single GEO call, and the standing rule says to run one archive query before bounding any absence by literature alone.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
Thirty-two queries run 2026-09-24 through scripts/pubmed_safe.py with zero instrument failures return 0 PubMed records for AFF1-AS1 under seven constructed names, Europe PMC full-text returns hitCount 0 for the quoted phrase, and GEO db=gds returns quotedphrasesnotfound for it, while positive controls through the identical shapes return 226 (AFF1), 1219 (AF4), 99 (HOXA-AS2), 633 (HOTAIR AND antisense), 651 (Europe PMC HOXA-AS2), 2129 (Europe PMC AFF1) and 24 (GEO HOXA-AS2).
Q155 bounded literature and archive search, with querytranslation recorded beside every count · Iqbal Lab oncofusion knowledge graph, beat Q155 · this project 2026What this evidence cannot show
This is a bound on a NAME LIST (AFF1-AS1, AFF1-AS, 'AFF1 antisense', 'AFF1 antisense RNA 1', LOC100506746, 'AF4 antisense', 'MLLT2 antisense') and on three indexes, not on the biology. Two entries in the script's 'control' group - "IRX2-DT"[tiab] and IRX2[tiab] AND antisense[tiab] - returned zero and were MISLABELLED by me: they are comparators, not controls, because they can legitimately be zero. The script's controls_all_positive flag therefore reads False, and the genuine controls listed in the claim are the ones that license the zeros. A control that can legitimately be zero is not a control.
Evidence that argues against this (1)
- The AFF1 symbol collides with AFF1 in Caenorhabditis elegans (Anchor-cell Fusion Failure 1, a eukaryotic fusogen), so AFF1-keyed counts are contaminated at an unmeasured rate and the 226 positive control is an upper bound on human-AFF1 records. This does not affect the zeros, which cannot be inflated by contamination, but it does mean the control's magnitude is overstated. Avinoam O et al. 2011
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.review
The 2026 Int J Cancer review of non-coding RNAs in KMT2A-rearranged acute lymphoblastic leukemia names AFF1 25 times and AFF1-AS1 zero times in its full text, and the lncRNA symbols it discusses are HOXA10-AS, LAMP5-AS1, LINC00226, LINC00511, LINC00958, LINC01221 and LINC01226.
The hidden regulators: Non-coding RNAs in KMT2A-rearranged acute lymphoblastic leukemia. · Poersch MA, Rodrigues AC, Stricker PEF, Azevedo ALK, Bruschi DP, de Oliveira JC · Int J Cancer 2026What this evidence cannot show
Cited for what it does NOT contain, which requires the instrument to be stated: Europe PMC fullTextXML for PMC12922646, HTTP 200, 176,836 bytes, <body> element present, 56,262 characters after stripping tags first and unescaping entities second, with U+2010/2011/2012/2013/2014/2212 all normalised to ASCII hyphen before searching. Without that normalisation the search for 'AFF1-AS1' would have returned zero for the wrong reason, because the document uses U+2010 throughout. A review's silence bounds the field's attention, not the literature; E1 bounds the literature.
Evidence that argues against this (1)
- The review also does not name C4orf36, which IS a protein-coding gene with a current-RefSeq TSS 269 bp from AFF1's, so its silence is about the whole 4q21.3 neighbourhood rather than about antisense RNAs specifically. Inferring that AFF1-AS1 in particular was considered and dismissed is not supported. UCSC Genome Browser et al. 2026
the bounded absence behind the gene entry: 32 queries, 3 indexes, and the 2026 topic review's full text
nobody has asked whether antisense transcription at the AFF1 locus is regulated, in any system