search

openforcefield / protein-ligand-benchmark

Protein-Ligand Benchmark Dataset for Free Energy Calculations
MIT License
144 stars 15 forks source link

Manually curated edges by Ken Takaba #98

Closed ijpulidos closed 1 year ago

ijpulidos commented 1 year ago

Adding manually curated edges by @kntkb

These changes add manually "optimized" edges for four targets, namely cdk2, mcl1, p38 and tyk2, which have shown to give good results when performing free energy calculations.

Additional information can be found in https://github.com/kntkb/protein-ligand-benchmark-custom/

ijpulidos commented 1 year ago

@IAlibay @hannahbaumann The style of the yaml file is not the same as in the others, I'd appreciate if you can share how you generated the other styles. I'm using plain pyyaml here and I don't think I can mix flow and block styles in the same file programatically. I do think that for the atom mappings the flow style is more readable.

ijpulidos commented 1 year ago

Script for preparation and formatting is in https://github.com/openforcefield/protein-ligand-benchmark/blob/new-perses-edges/preparation/mappers/perses-mapper.py (thanks @IAlibay for pointing it out). I was missing the default_flow_style=None. This should be ready for review now.

ijpulidos commented 1 year ago

Oh, I just realized this was merged. I made some new changes in the same branch, correcting the issues that @hannahbaumann was having with them. I think I'll be adding the new changes in the base branch/PR that we are using for this, without having to open a new PR. I hope that's okay.

IAlibay commented 1 year ago

Oh, I just realized this was merged. I made some new changes in the same branch, correcting the issues that @hannahbaumann was having with them. I think I'll be adding the new changes in the base branch/PR that we are using for this, without having to open a new PR. I hope that's okay.

Please do!