Open dan2097 opened 6 years ago
I know it's a bit outdated, @dan2097 could you double check it on the fresh release?
@MysterionRise Unfortunately not really as a few months after this we switched to using Bingo NoSQL. I think as I haven't provided sufficient information to reproduce the issue from scratch that this issue should be closed.
I'm using a version built from source on the 16th November 2017 against PostgreSQL 9.6.
select count(*) from bingo.chemdata where smiles @ ('0.95', '1', 'S(C1C=CC(=C(C2=NC(C3=C(C(CCC)=NN3C)N2)=O)C=1)OCC)(N1CCN(C)CC1)(=O)=O', 'Tanimoto')::bingo.sim
9select count(*) from bingo.chemdata where smiles @ ('0.95', '1.01', 'S(C1C=CC(=C(C2=NC(C3=C(C(CCC)=NN3C)N2)=O)C=1)OCC)(N1CCN(C)CC1)(=O)=O', 'Tanimoto')::bingo.sim
14select count(*) from bingo.chemdata where smiles @ ('0.95', '1.02', 'S(C1C=CC(=C(C2=NC(C3=C(C(CCC)=NN3C)N2)=O)C=1)OCC)(N1CCN(C)CC1)(=O)=O', 'Tanimoto')::bingo.sim
17 [the identical compound had a similarity somewhere between 1.01 and 1.02]This behaviour did not exist in a version from ~2014.
Another minor bug is that if the upper bound on similarity is given as a null, this is interpreted as 0 rather than 1 (or perhaps a number greater than 1!) and hence the query is rejected.