Oncofusion

start / graph / ana-q199-aff1-holds-one-dense-201-kb-germline-signal-block-dominated-by-lipid-traits-at-the-99th-length-matched-percentile

Analysismoderate confidence 3 of 5: One strong study, or several smaller consistent ones.

AFF1, the fusion's 3-prime partner, carries 126 genome-wide-significant in-gene SNPs across 505 traits in one contiguous 201 kb block (98.9th length-matched percentile), 631 of 849 rows lipid-related and 68 red-cell indices

The gene that supplies half of the leukaemia fusion protein also sits in one stretch of inherited DNA that influences hundreds of blood-fat and red-cell measurements. Most of that is probably one signal repeated, so it is a lead and not a finding.

The same thing, for a scientist

In-gene genome-wide-significant variants only (intergenic excluded). AFF1: 126 distinct SNPs, 849 association rows, 505 traits, all inside chr4:86939360-87140201 (201 kb). Length-matched comparison against 1461 protein-coding genes of 0.7-1.4x its length: more in-gene SNPs than 98.9% of them. Trait mix: lipid/fatty-acid rows 631, red-cell-index rows 68; top traits Triglyceride levels (17); Triglycerides (11); Mean corpuscular hemoglobin (11); Mean corpuscular volume (10).

Go deeper
Why 126 SNPs are not 126 signals

All of them lie in one ~200 kb stretch, so they are in strong linkage disequilibrium and mostly tag the same underlying variant. The 505 traits are largely correlated metabolite and lipid measures from large biobank panels, which multiplies rows. The honest unit is roughly one signal block, whose strength (98.9th percentile among length-matched genes) is real but whose independent count was not tested.

A column that was dropped

An 'independent loci' count that merged SNPs within 100 kb scored every gene shorter than 100 kb as exactly 1, which is meaningless. It is not reported.

Facts and figures
scriptanalyses/q199-germline-genetics-of-the-fusion-pathway/run.py
aff1
{
  "snps": 126,
  "rows": 849,
  "traits": 505,
  "span_chr4_bp": [
    86939360,
    87140201
  ],
  "lipid_rows": 631,
  "red_cell_rows": 68,
  "top_traits": {
    "Triglyceride levels": 17,
    "Triglycerides": 11,
    "Mean corpuscular hemoglobin": 11,
    "Mean corpuscular volume": 10,
    "Hematocrit": 8,
    "Polyunsaturated fatty acids to monounsaturated fatty acids ratio": 8
  }
}
matched
{
  "gene": "AFF1",
  "len_kb": 206.9,
  "snps": 126,
  "rows": 849,
  "traits": 505,
  "n_length_matched": 1461,
  "pct_length_matched": 98.9,
  "context": {
    "intron_variant": 817,
    "3_prime_UTR_variant": 24
  }
}
How this could bite us back
  • A drug that lowers AFF1-family activity could, if the germline signal reflects real AFF1 function, move lipid or erythroid indices. This is a hypothesis to monitor, not a prediction.
Still open
  1. Is the AFF1 signal AFF1? Colocalisation of the 4q21 block with cis-eQTLs (GTEx, eQTLGen, blood) would say whether AFF1 expression, or a neighbouring gene, carries it - a computation suited to the cluster.
  2. How many conditionally independent signals does the block hold?
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    In the GWAS Catalog, AFF1 has 126 distinct genome-wide-significant in-gene SNPs (849 rows, 505 traits) within a single contiguous 201 kb span of chromosome 4, more than 98.9% of 1461 protein-coding genes of similar length.

    What this evidence cannot show

    In-gene = INTERGENIC flag 0 and the gene named in MAPPED_GENE. Gene lengths from Ensembl BioMart protein-coding genes; the matched set excludes AFF1 itself. Ties scored by mid-rank.

    Evidence that argues against this (1)
    • Row and SNP counts are inflated by linkage disequilibrium (one block) and by correlated metabolite traits; conditional-independence analysis was not run, so the number of independent signals is unknown and probably small. The Catalog maps a SNP to a gene by proximity, so a signal may belong to a neighbouring gene. GWAS Catalog 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Of AFF1's 849 in-gene genome-wide-significant rows, 631 are lipid or fatty-acid traits and 68 are red-cell indices (mean corpuscular volume, haemoglobin, haematocrit, red cell count), with triglycerides the most frequent single trait.

    What this evidence cannot show

    Trait families assigned by whole-word patterns on the trait string; a row can match more than one family (e.g. a lipid-to-fatty-acid ratio).

    Evidence that argues against this (1)
    • A germline regulatory signal in AFF1 says nothing direct about the SOMATIC KMT2A::AFF1 fusion. It suggests only that the normal gene has a role in lipid and erythroid biology, which would matter as an on-target consideration for any drug that reduces AFF1-family activity, and which is untested here. GWAS Catalog 2026
Connected to