How to open the OpenBind Zika virus blind challenge data in a Jupyter notebook
The 400 blinded compounds are in parquet/compounds.parquet, which opens with pyarrow, pandas or DuckDB. In a Scigantic notebook the challenge files and the OpenADMET tutorial are mounted at /mnt/archive, so there is nothing to pull from Hugging Face or GitHub.
The code on this page runs in a Scigantic hosted notebook, where the dataset is mounted read-only at /mnt/archive. The same path is in the SCIGANTIC_MOUNT_PATH environment variable and linked from ~/archive. The public source is s3://scigantic-openbind-challenge-1 (AWS us-east-1, public, no account needed).
Open OpenBind Zika challenge data on SciganticOpen the sample notebook: OpenBind Zika challenge tutorial, interim first place
Code
Read the compounds table
import os
import pyarrow.parquet as pq
pf = pq.ParquetFile(os.path.join(os.environ['SCIGANTIC_MOUNT_PATH'], 'parquet/compounds.parquet'))
print(pf.metadata)
df = pf.read_row_group(0).to_pandas()
df.head()
Run on 2026-11-16 in the Scigantic notebook image against the mounted data. It returned: Parquet 400 rows, 19 columns, 1 row groups; columns ['molecule_name:string', 'smiles:string', 'canonical_smiles:string', 'inchikey:string', 'molecular_formula:string', 'molecular_weight:double', 'heavy_atoms:int64', 'clogp:double', 'tpsa:double', 'hbd:int64']
Join compounds to the baseline structures with DuckDB
import os
from pathlib import Path
import duckdb
ARCHIVE = Path(os.environ.get("SCIGANTIC_MOUNT_PATH") or Path.home() / "archive")
con = duckdb.connect()
for name in ("compounds", "baseline_structures"):
con.execute(
f"CREATE OR REPLACE VIEW {name} AS "
f"SELECT * FROM read_parquet('{ARCHIVE / 'parquet' / (name + '.parquet')}')"
)
# 400 blinded compounds; the Boltz-2 baseline misses the pocket on 21 of them.
con.sql("""
SELECT count(*) AS compounds,
count(distinct murcko_scaffold) AS scaffolds,
count(*) FILTER (WHERE NOT b.in_consensus_pocket) AS baseline_off_pocket,
round(median(b.pocket_offset_a), 2) AS median_pocket_offset_a
FROM compounds c JOIN baseline_structures b USING (molecule_name)
""").df()
The starter cell from this dataset's schema card. It ran without error on 2026-11-16 in the Scigantic notebook image against the mounted data.
What the files are
OpenBind Blind Challenge 1 is a structure-prediction competition on the Zika virus NS2B-NS3 protease. 400 compounds were soaked into crystals and solved; participants get only the SMILES and predict each bound complex.
This archive holds the blinded compound list, the OpenADMET tutorial repository at commit 87acddb with its scoring code and submission validator, an example submission of 400 Boltz-2 predictions, and two reference structures. The experimental structures are withheld until the challenge closes.
How the files are laid out
| Path under the mount | What it holds |
|---|---|
| compounds/openbind-blind-challenge-1-TEST-BLINDED.csv | 400 rows, two columns: Molecule_Name and SMILES. This is the whole blinded input. |
| parquet/compounds.parquet | The same 400 compounds with RDKit descriptors, InChIKeys and Murcko scaffolds precomputed (19 columns). |
| parquet/baseline_structures.parquet | 400 rows, 12.3 MB: the full PDB text of each baseline complex plus per-pose geometry. |
| tutorial/ | The OpenADMET tutorial repository: notebooks, inputs, the official scorer in evaluation/, the submission validator in validation/. |
| tutorial/outputs/example_structure_submission/<id>.pdb | The example submission, stored unzipped as 400 PDB files of about 135 KB each. |
| reference-structures/ | RCSB entries 8PN6 (apo) and 29HN (the closed conformation used in the challenge). |
Size, format and license
| Format | Parquet, CSV, PDB and mmCIF. Nothing in the bucket is compressed. |
|---|---|
| Size | 428 objects, 68.3 MB |
| License | Apache 2.0 |
| Provider | OpenADMET and the OpenBind Consortium; mirrored by Scigantic |
| Public source | s3://scigantic-openbind-challenge-1 (AWS us-east-1, public, no account needed) |
| Notebook image | Cryo-EM and structural biology (RDKit, gemmi, DuckDB and pyarrow preinstalled) |
Things to know before you read it
- Nothing here can score a prediction: the experimental structures are not public while the challenge is open.
- Read many baseline structures from
parquet/baseline_structures.parquet. One read replaces 400 separate file reads.
Run it
Start a notebook from the OpenBind Zika challenge data archive page. The data is mounted when the notebook opens, so the code above runs as written and nothing is downloaded first. For a longer worked example, open the sample notebook OpenBind Zika challenge tutorial, interim first place.
Related guides
- How to open the Alchemistry 2027 dengue binding free energy challenge data in a Jupyter notebook
- How to open the OpenADMET CYP inhibition challenge data in a Jupyter notebook
- How to query ChEMBL bioactivity data as Parquet without installing the database
- How to query 122 million PubChem compounds with ADMET predictions in a Jupyter notebook
- How to read prebuilt MSA databases for structure prediction (Thal-Kak) without unpacking them
Last checked 2026-11-21. All dataset guides.