search

MDAnalysis / RDKitConverter-benchmark

Benchmark for the RDKitConverter's inferring of bond orders and charges
GNU General Public License v2.0
3 stars 0 forks source link

Benchmark for MDA 2.4.3 #5

Closed cbouy closed 2 months ago

cbouy commented 1 year ago

Fixes #4

I've updated the scripts to run on the latest ChEMBL release (ChEMBL33), updated RDKit to 2022.09.1 and python to 3.10.11 and reran the entire thing for MDAnalysis 2.4.3 as it was the latest release before my modifications.

I'll try to rerun it this week for MDA 2.5.0 as there are some changes in there which should improve the accuracy.

cbouy commented 1 year ago

@IAlibay @richardjgowers FYI I've updated and ran the RDKitConverter accuracy benchmark for MDAnalysis 2.4.3, next I'll do 2.5.0

cbouy commented 1 year ago

I added a scaffold network analysis to extract which scaffolds are problematic in structures during bond order inferring, so that one day I can try to properly fix it...

cbouy commented 2 months ago

Oh my this is more than a year old... I'll merge it in as it contains some interesting tables to better understand which substructures are prone to failing and hopefully at some point I can rerun it for 2.5.0