The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
foldserve benchmark inputs
The multiple sequence alignments of foldserve's speed tests, the ones its published time tables were measured with.
ladder/<N>/msa/<k>_<PDB>_<chain>.a3m.gz is one chain's alignment for the <N>-token input
benchmarks/inputs/ladder/<N>/<k>_<PDB>.json of the foldserve repository (sizes 200, 400, 600 and 800,
and 1024, 1536 and 2048 for the large sizes). SHA256SUMS lists every file's digest.
Fetch them into a clone of foldserve and unpack them where the inputs name them.
hf download clorf6/foldserve-benchmarks --repo-type dataset --include "ladder/*" --local-dir benchmarks/inputs
gunzip -k benchmarks/inputs/ladder/*/msa/*.a3m.gz
benchmarks/inputs/ladder/build.py in the repository made them. Each chain was searched on
ColabFold's MMseqs2 server (UniRef30 and ColabFold's environmental databases), and its alignment was
cropped to the chain's trimmed segment. A new search can return different rows, because the server's
databases change over time.
Credit
The alignments come from the ColabFold MMseqs2 server. Please cite ColabFold (Mirdita et al., Nature Methods 19, 679-682, 2022) and MMseqs2 (Steinegger and Söding, Nature Biotechnology 35, 1026-1028, 2017), and follow the terms of the databases they search (UniRef30 from UniProt, CC BY 4.0, and ColabFold's environmental databases).
- Downloads last month
- 22