ChunkedEncodingError: search may be incomplete

[Moh-Pou]

I am getting an error when searching for aspirin in REAL, indicating that it doesn't search in the entire REAL so the code can't find Aspirin:

Python 3.8.16 | packaged by conda-forge | (default, Feb 1 2023, 16:01:55)
[GCC 11.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.

import pandas as pd
from smallworld_api import SmallWorld
aspirin = 'O=C(C)Oc1ccccc1C(=O)O'
sw = SmallWorld()
results : pd.DataFrame = sw.search(aspirin, dist=0, db=sw.REAL_dataset)
ChunkedEncodingError: search may be incomplete
results['dist'].values
array([2, 2, 2, 2, 2, 3, 3, 3, 3, 3])

[David-Araripe]

This seems to be a problem on Small World's side, you can check a previous discussion about it in the previous issue #2

[matteoferla]

The chunking error in SW (not in Arthor) is a concern.

But the aspirin in REAL is weird. And its oddity spills over to Arthor. I have never cared too much to actually enquire or dig deep into.
I first I had assumed Lipinski rules don't play well with it and then I was told that REAL BB is allegedly not a subset of REAL DB, a misconception I held for a long time.

However, Aspirin (ZINC0053) is in Enamine REAL DB (Z104474430) and Enamine BB (EN300-19606 —$24/g 😆). The e-store link doesn't work as of July 23, but it ought to have.

So my guess is that the fingerprint heuristics struggle to find such an unremarkable compound, that is just a benzene scaffold with a just a carboxylate and a ester. Going one step further, benzoate (O=C(O)c1ccccc1) in Arthor against BB works but it shows what's going on under the hood, with a deluge of bezenyl and carboxyl compounds of diverse sizes...