Lesson 18 of 20 · 10 min

MANE: One Gene, One Agreed Transcript

By the end of this lesson you can look at two accessions from two different agencies — one starting NM_, one starting ENST — and know, without checking, that they describe the exact same transcript: same exons, same coding sequence, same protein. That is not a coincidence. It is the deliverable of a project built to end it.

The problem MANE was built to fix

In the ID mapping lesson you saw that a RefSeq ID and an Ensembl ID for "the same gene" can still point at different transcript models — different exon boundaries, a different chosen start codon, a slightly different protein. Two independent annotation pipelines, run by two independent teams, rarely land on the identical structure by chance. For most of genomics history that meant every clinical lab, every paper, and every variant caller had to pick a side and hope the other side's tools agreed.

MANE — Matched Annotation from NCBI and EMBL-EBI — is a joint project between the RefSeq team at NCBI and the Ensembl/GENCODE team at EMBL-EBI that fixes this by agreement rather than by chance. For each human protein-coding gene, curators from both institutes sit down and pick one transcript each, then check that the two picks are sequence-identical: same exon coordinates on the genome, same CDS, same translated protein. Where they don't match, the teams revise annotation until they do. The result is a db_xref-style link (covered in the db_xref lesson) that for once is a guarantee, not just a pointer.

MANE Select vs MANE Plus Clinical

MANE ships two tiers of matched transcript, and the difference matters for how you use them.

	MANE Select	MANE Plus Clinical
Purpose	The one representative transcript per gene	Extra transcript(s) where Select alone misses known clinical variants
How many genes	Nearly every human protein-coding gene	Several dozen genes, where Select drops an exon carrying a reported pathogenic variant
Count per gene	Exactly one	Zero, one, or occasionally more
Typical use	Default transcript for annotation, reporting, tooling	Supplement — check both when Select doesn't cover a known variant

MANE Select is the default: pick a gene, get one transcript, and it is the same transcript whether you fetched it from RefSeq or Ensembl. MANE Plus Clinical exists because several dozen genes have well-documented pathogenic variants that fall in an exon the Select transcript doesn't include — so MANE adds a second matched pair just for that locus, rather than forcing Select to serve two masters.

BRCA1: the case study

BRCA1 (NCBI Gene 672, HGNC HGNC:1100) is MANE's textbook example. Its RefSeq mRNA NM_007294.4 is the MANE Select transcript, and it is matched to Ensembl transcript ENST00000357654.9, whose protein is ENSP00000350283.3 — the same protein NCBI calls NP_009225.1. Fetch the GenBank flat file for NM_007294 and the record tells you this directly, in its own words:

python

from Bio import Entrez, SeqIO

Entrez.email = "[email protected]"
handle = Entrez.efetch(db="nuccore", id="NM_007294", rettype="gb", retmode="text")
record = SeqIO.read(handle, "genbank")
handle.close()

print(record.annotations["keywords"])
# ['RefSeq', 'MANE Select']

for line in record.annotations["comment"].splitlines():
    if "MANE Ensembl match" in line:
        print(line.strip())
# MANE Ensembl match     :: ENST00000357654.9/ ENSP00000350283.3

That MANE Ensembl match line is the record telling you, in its own annotation, exactly which Ensembl transcript and protein are guaranteed identical to this one. No BLAST, no manual alignment, no trust required — it's a stated fact in the record.

What "identical" does and doesn't cover

Here is the scope you must keep straight: MANE's identity guarantee applies only to the matched pair, not to every RefSeq or Ensembl ID that happens to touch the same gene. NM_007294.4 and ENST00000357654.9 are interchangeable because MANE says so, explicitly, for that pair. BRCA1's NCBI Gene 672 record lists dozens of other RefSeq transcript variants — none of them carries a MANE Ensembl match line, and none of them is guaranteed to match any particular Ensembl transcript. Likewise, ENSG00000012048 (the Ensembl gene ID) can have many transcripts beyond the MANE-matched one, with their own exon structures.

Gotcha: "RefSeq and Ensembl agree on BRCA1" is not the same claim as "every RefSeq BRCA1 transcript equals some Ensembl BRCA1 transcript." MANE's synonymy is a property of the MANE Select or MANE Plus Clinical pair specifically — check for the MANE Ensembl match line (or the equivalent MANE_Select tag on the Ensembl side) before assuming any other transcript pair lines up.

Why this ends a decade of friction

As of the most recent MANE release, MANE Select covers over 97% of human protein-coding genes — essentially the whole coding genome — with each gene's Select transcript sequence-identical between RefSeq and the Ensembl annotation GENCODE also distributes. That is why ClinGen and ACMG variant-classification guidance now recommend reporting clinical variants against the MANE Select transcript by default: a coordinate on NM_007294.4 and the same coordinate on ENST00000357654.9 describe the identical base, so a lab using RefSeq tools and a lab using Ensembl tools finally produce comparable reports. It is the single clearest case of two independent database ecosystems choosing interoperability over territory — and it is why, from here on, you can treat a gene's MANE Select pair as one ID with two spellings.

ID Mapping in Practice Genome Assemblies and Builds