Dataset Viewer
Auto-converted to Parquet Duplicate
id
large_stringlengths
12
35
length
int64
55
845
offset
int64
0
2.31M
truncated
bool
1 class
ATP6|acinonyx-jubatus
226
0
false
ATP6|aethalodelphis-obliquidens
226
226
false
ATP6|ailuropoda-melanoleuca
226
452
false
ATP6|antechinus-flavipes
226
678
false
ATP6|apodemus-sylvaticus
226
904
false
ATP6|artibeus-jamaicensis
226
1,130
false
ATP6|balaenoptera-acutorostrata
226
1,356
false
ATP6|balaenoptera-borealis
226
1,582
false
ATP6|balaenoptera-musculus
226
1,808
false
ATP6|balaenoptera-physalus
226
2,034
false
ATP6|bison-bison
226
2,260
false
ATP6|bos-grunniens
226
2,486
false
ATP6|bos-javanicus
226
2,712
false
ATP6|bos-taurus
226
2,938
false
ATP6|bubalus-bubalis
226
3,164
false
ATP6|budorcas-taxicolor
226
3,390
false
ATP6|callithrix-jacchus
226
3,616
false
ATP6|callospermophilus-lateralis
226
3,842
false
ATP6|camelus-bactrianus
226
4,068
false
ATP6|camelus-dromedarius
226
4,294
false
ATP6|canis-aureus
226
4,520
false
ATP6|canis-lupus
226
4,746
false
ATP6|capra-hircus
226
4,972
false
ATP6|capricornis-sumatraensis
226
5,198
false
ATP6|castor-canadensis
226
5,424
false
ATP6|cavia-porcellus
226
5,650
false
ATP6|cebus-apella
226
5,876
false
ATP6|ceratotherium-simum
226
6,102
false
ATP6|cervus-elaphus
226
6,328
false
ATP6|chinchilla-lanigera
226
6,554
false
ATP6|choloepus-didactylus
226
6,780
false
ATP6|cnephaeus-nilssonii
226
7,006
false
ATP6|condylura-cristata
226
7,232
false
ATP6|crocuta-crocuta
226
7,458
false
ATP6|dama-dama
226
7,684
false
ATP6|dasypus-novemcinctus
226
7,910
false
ATP6|delphinapterus-leucas
226
8,136
false
ATP6|delphinus-delphis
226
8,362
false
ATP6|desmodus-rotundus
226
8,588
false
ATP6|diceros-bicornis
226
8,814
false
ATP6|dipodomys-merriami
226
9,040
false
ATP6|dipodomys-spectabilis
226
9,266
false
ATP6|echinops-telfairi
226
9,492
false
ATP6|elephas-maximus
222
9,718
false
ATP6|enhydra-lutris
226
9,940
false
ATP6|equus-asinus
226
10,166
false
ATP6|equus-caballus
226
10,392
false
ATP6|equus-quagga
226
10,618
false
ATP6|erinaceus-europaeus
226
10,844
false
ATP6|eschrichtius-robustus
226
11,070
false
ATP6|eubalaena-glacialis
226
11,296
false
ATP6|eumetopias-jubatus
226
11,522
false
ATP6|felis-catus
226
11,748
false
ATP6|fukomys-damarensis
226
11,974
false
ATP6|globicephala-melas
226
12,200
false
ATP6|gorilla-gorilla
226
12,426
false
ATP6|gracilinanus-agilis
226
12,652
false
ATP6|grampus-griseus
226
12,878
false
ATP6|halichoerus-grypus
226
13,104
false
ATP6|herpailurus-yagouaroundi
226
13,330
false
ATP6|heterocephalus-glaber
226
13,556
false
ATP6|hippopotamus-amphibius
226
13,782
false
ATP6|homo-sapiens
226
14,008
false
ATP6|hyaena-hyaena
226
14,234
false
ATP6|hyperoodon-ampullatus
226
14,460
false
ATP6|ictidomys-tridecemlineatus
226
14,686
false
ATP6|jaculus-jaculus
226
14,912
false
ATP6|kogia-breviceps
226
15,138
false
ATP6|lama-pacos
226
15,364
false
ATP6|lemur-catta
226
15,590
false
ATP6|leopardus-geoffroyi
226
15,816
false
ATP6|lepus-europaeus
226
16,042
false
ATP6|leucopleurus-acutus
226
16,268
false
ATP6|lipotes-vexillifer
226
16,494
false
ATP6|lontra-canadensis
226
16,720
false
ATP6|loxodonta-africana
222
16,946
false
ATP6|lutra-lutra
226
17,168
false
ATP6|lynx-canadensis
226
17,394
false
ATP6|lynx-rufus
226
17,620
false
ATP6|macaca-fascicularis
226
17,846
false
ATP6|macaca-mulatta
226
18,072
false
ATP6|macaca-nemestrina
226
18,298
false
ATP6|macaca-thibetana
226
18,524
false
ATP6|macrotis-lagotis
226
18,750
false
ATP6|mandrillus-leucophaeus
226
18,976
false
ATP6|marmota-flaviventris
226
19,202
false
ATP6|marmota-monax
226
19,428
false
ATP6|martes-martes
226
19,654
false
ATP6|meles-meles
226
19,880
false
ATP6|meriones-unguiculatus
226
20,106
false
ATP6|mesocricetus-auratus
226
20,332
false
ATP6|microcebus-murinus
226
20,558
false
ATP6|micromys-minutus
226
20,784
false
ATP6|microtus-ochrogaster
226
21,010
false
ATP6|molossus-molossus
226
21,236
false
ATP6|monodelphis-domestica
226
21,462
false
ATP6|monodon-monoceros
226
21,688
false
ATP6|muntiacus-reevesi
226
21,914
false
ATP6|mus-musculus
226
22,140
false
ATP6|muscardinus-avellanarius
226
22,366
false
End of preview. Expand in Data Studio

MitoAge: ESM-2 embeddings of mitochondrial proteins and their nuclear partners

Protein language model embeddings (ESM-2 650M, esm2_t33_650M_UR50D, last layer, float16; per residue and their mean over residues) of the 13 mitochondrially encoded proteins and of 62 nuclear proteins in up to 184 mammals, for the test of whether partners' embeddings move together across species. Computed in the kron-fm project (its script esm_embed.py).

Part of the prepared data of dna-seq/mitoage-pipelines, the analysis code of MitoAge. The code, the hypotheses and the results are described there (README.md, docs/HYPOTHESES.md, docs/PIPELINE.md). Fetch it with uv run mitoage download prepared --topic embeddings in that repository: the folders land in its data/.

Sister datasets: the MitoAge build, nuclear partners, embeddings, aligner variants, structures, results.

Contents

Folder What Made by
input/embeddings/nuclear-esm2_650m.{index.parquet,mean.npy} 10,639 nuclear proteins; the index gives id (GENE|species-slug), length, offset; mean.npy is [n, 1280] in index order kron-fm
input/embeddings/mt13-esm2_650m.{index.parquet,mean.npy} the 13 mitochondrial proteins of 184 mammals, 2,392 proteins kron-fm
input/embeddings/nuclear-esm2_650m.residues.npy per-residue embeddings of the same proteins, [2,308,802, 1280] float16 (5.9 GB); a protein's rows are offset to offset + length of the index kron-fm
input/embeddings/mt13-esm2_650m.residues.npy per-residue embeddings of the mitochondrial proteins, [697,368, 1280] float16 (1.8 GB) kron-fm
input/embeddings/nuclear.faa the nuclear sequences that were embedded mitoage mitoport select

The per-residue files are what mitoage mitoport iface reads for its esm_ columns (the coupling of complex IV's contact residues, H5 in docs/HYPOTHESES.md); mitoage download prepared --topic residues fetches them, and --topic embeddings leaves them out. Added on 2026-11-15, as the files the published interface run read; the index and mean files are unchanged and the mean of each protein's residue rows equals its row of mean.npy to float16 rounding (at most 0.0024 over 400 proteins checked). SHA-256:

76c9a99543838ae326ab19c0ce0167c86f645a1573cbc2d3230263bdec133954  mt13-esm2_650m.residues.npy
d60ef0ecabc24eeea0ba36375637eee277ef1a6f6f8fc24efbf6cd637443b197  nuclear-esm2_650m.residues.npy

Nothing here is peer reviewed; docs/HYPOTHESES.md in the code repository says which results held.

Licences

Sequences: NCBI, public domain, and the MitoAge build (see its card). Embeddings: CC BY 4.0. ESM-2 is released by Meta under the MIT licence.

Downloads last month
63

Collection including longevity-genie/mitoage-embeddings