ACER1 is on the list of in-frame reciprocal KMT2A partners - and ACER1 is the gene the 'reciprocal-negative' breakpoints fall next to

In a spliced KMT2A::MLLT1 the break falls in the intergenic stretch between ACER1 and MLLT1 on 19p13.3, so der(11) receives an intact MLLT1 and der(19) receives 3'-KMT2A abutting the ACER1 side of the break 1. If the break falls in the intergenic DNA, der(19) carries no coding sequence to fuse to and there is no reciprocal gene - the standard account 3. If it falls a little further 5', inside ACER1, then der(19) carries ACER1 exons joined to 3'-KMT2A, and if the frames line up an ACER1::KMT2A chimera results. That is a difference of a few kilobases in breakpoint position with a qualitative consequence, and it is the kind of thing a breakpoint class defined by 'upstream of exon 1' cannot see.

Firm on the fact, soft on the attribution. Fact: ACER1 is in the alphabetical list of 40 in-frame fused reciprocal TPGs in Supplementary Table S4A, read from the supplementary Word file of the recombinome, and it carries no '(2x)' marker so it is a single case 2. Fact: the direct partners of the complex rearrangements that produced those 40 in-frame reciprocal fusions are listed, and MLLT1 is among them 2. Not established: that the ACER1 case is one of the MLLT1 cases. The supplementary table is a flat alphabetical gene list with no patient-level linkage, so the connection is made here by genomic adjacency - ACER1 sits immediately 5' of MLLT1 at 19p13.3, and a break producing an in-frame ACER1::KMT2A reciprocal in any other context would require a separate translocation to 19p13.3 that is not the one under discussion. That is a strong argument and it is not a citation. One line from the source archive settles it and nothing else will.

It does not dissolve the arm. One case against 153 is a contamination rate under 1% if the attribution is right, and any residual reciprocal-positive patients sitting inside the reciprocal-negative arm dilute a true effect rather than create a false one. What it does is remove the word 'by construction'. The graph has been saying that a spliced fusion produces no reciprocal allele as a matter of geometry rather than of measurement, and that is not quite right: it produces no reciprocal allele provided the break lands in the intergenic DNA rather than in the upstream gene. The correct statement is probabilistic and the probability has not been measured. Pair this with the converse error - patients with intra-MLLT1 breaks who are nonetheless reciprocal-negative because 3'-KMT2A was deleted 4 - and both arms of the comparison leak, in opposite directions, by unmeasured amounts.

This finding is not in the paper's text, its figures or its abstract. It is one gene name in an alphabetical list inside a 425 KB Word file attached to the supplementary archive, and it contradicts a sentence in the paper's own Results. It was found by fetching {PMCID}/supplementaryFiles, checking that the .docx began with the PK magic bytes rather than trusting the HTTP 200, and flattening word/document.xml to text. The archive is 1,027,204 bytes and the Word file 425,197 bytes uncompressed - both real, unlike the zero-byte member this project logged for PMC6880934.