Data & methods
Filo3D is assembled by a reproducible pipeline (Pixi + DVC), not by hand. See the GitHub repository for the code.
Acquisition — two routes
- NCBI taxonomy (trusted core). An RCSB search on source-organism lineage =
Filoviridae (taxid
11266) returns every experimental polymer entity below the family; metadata is fetched from the RCSB Data API. Entities are classified into canonical proteins and species, then CIFs are downloaded and annotated with wwPDB validation percentiles. - Sequence-similarity backstop. An RCSB mmseqs2 search against tag-free reference
sequences catches structures the taxonomy route misses (e.g. host-annotated or
synthetic constructs, such as the bat Myotis VP35
6DKU). These are written to a review queue, never auto-added.
Taxonomy backbone follows Carroll et al. 2013.
Composition (polymer entities)
| Protein | BDBV | EBOV | LLOV | MARV | MLAV | RAVV | RESTV | SUDV | TAFV | All |
|---|---|---|---|---|---|---|---|---|---|---|
| GP | 9 | 132 | 0 | 9 | 2 | 10 | 0 | 12 | 0 | 174 |
| NP | 2 | 16 | 2 | 8 | 7 | 0 | 1 | 3 | 1 | 40 |
| VP35 | 0 | 34 | 0 | 10 | 0 | 0 | 6 | 2 | 0 | 52 |
| VP40 | 1 | 13 | 0 | 1 | 0 | 0 | 0 | 18 | 0 | 33 |
| VP30 | 0 | 9 | 1 | 1 | 2 | 0 | 1 | 0 | 0 | 14 |
| VP24 | 0 | 6 | 0 | 1 | 0 | 0 | 1 | 2 | 0 | 10 |
| L | 0 | 7 | 0 | 2 | 0 | 0 | 0 | 1 | 0 | 10 |
| All | 12 | 217 | 3 | 32 | 11 | 10 | 9 | 38 | 1 | 333 |
Notes
- Species codes: EBOV (Zaire), SUDV (Sudan), BDBV (Bundibugyo), RESTV (Reston), TAFV (Taï Forest), MARV (Marburg), RAVV (Ravn), LLOV (Lloviu), MLAV (Měnglà).
- One row per polymer entity; a structure may contribute several (e.g. GP1 + GP2, or a bound Fab). Sequences carry expression tags as deposited.
- Amino-acid sequences and the full master table are in the repository
(
data/aa/fa/filo3d.fa,data/metadata/clean.csv).