CoolFace
Datasetpublic

marin-dna/animals-cds-proj-v1-all204

Animal CDS via human→animal projection — all animal orders Human protein-coding (CDS) 255 bp windows projected onto 204 animal species (one annotated genome per Metazoan order) by mmseqs2 nucleotide local alignment — the projection ("Arm B") side of the issue #353 CDS projection-vs-annotation experiment. How it was built The human v5 CDS intervals (Ensembl coding exons, filtered 20–10,000 bp, +20 bp splice flank, expanded ≥256 bp) are tiled into 255 bp windows… See the full description on the dataset page: https://huggingface.co/datasets/marin-dna/animals-cds-proj-v1-all204.

sourceHugging Faceupdated 3mo agoView on Hugging Face
0likes358downloads
Dataset Card

Animal CDS via human→animal projection — all animal orders

Human protein-coding (CDS) 255 bp windows projected onto 204 animal species (one annotated genome per Metazoan order) by mmseqs2 nucleotide local alignment — the projection ("Arm B") side of the issue #353 CDS projection-vs-annotation experiment.

How it was built

The human v5 CDS intervals (Ensembl coding exons, filtered 20–10,000 bp, +20 bp splice flank, expanded ≥256 bp) are tiled into 255 bp windows (269,866 windows) and searched against each target genome with mmseqs2 (--search-type 3 --strand 2 --mask-lower-case 1 -s 7.5 --max-accept 1 — the v30 operating point). The best hit per window (by bitscore) is kept and midpoint-resized to exactly 255 bp. Pipeline: `cds_projection.smk`; species list: `animals_order204.tsv`.

⚠️ Coverage is vertebrate-weighted

Nucleotide CDS projection reaches vertebrates well but fades across the animal tree: 97% of the ~12.6 M retained windows come from the 125 chordates (mammals ~79%, birds/reptiles ~37%, fish ~20%, jawless ~8–10% of human CDS windows projected); the 79 invertebrate species (16 phyla) contribute only ~3.7%, at a ~2% reach floor. A vertebrate-only subset is at bolinas-dna/animals-cds-proj-v1-vert125.

Schema

columntypedescription
idstrtarget locus chrom:start-end (0-based half-open, 255 bp)
seqstr255 bp target-genome sequence (+ strand), reverse-complement-augmented

24,319,966 rows (12,159,983 projected windows × 2 for RC augmentation), 204 species.