Add Sage 2.2.1 for lipids v1.0 dataset benchmarking

.github/workflows/opt.yaml

@@ -146,7 +146,7 @@ jobs:
 
           input_file=${{ inputs.path }}                        # path to the input YAML file
           input_dir=$(dirname $input_file)                     # parent directory of input YAML file
-          git add $input_dir/output/{dde,icrmsd,rmsd,tfd}.csv
+          git add $input_dir/output/{dde,rmsd,tfd}.csv
           git add $input_dir/output/{dde,rmsd,rmsd_cdf,tfd,tfd_cdf,bonds,angles,dihedrals,impropers}.png
           git commit -m "Add benchmark results"
           git push
@@ -160,7 +160,7 @@ jobs:
           micromamba env export > env.yaml
           input_files=$(python get_files.py ${{ inputs.path }})
           tar cf results.tar \
-            tmp.sqlite.bz2 $input_dir/output/{dde,icrmsd,rmsd,tfd}.csv \
+            tmp.sqlite.bz2 $input_dir/output/{dde,rmsd,tfd}.csv \
             $input_dir/output/{dde,rmsd,rmsd_cdf,tfd,tfd_cdf,bonds,angles,dihedrals,impropers}.png \
             env.yaml main.py $input_files
 
@@ -179,7 +179,7 @@ jobs:
         run: |
           input_files=$(python get_files.py ${{ inputs.path }})
           python zenodo_upload.py --title "YDS Upload ${{ inputs.pr_number }}" \
-            tmp.sqlite.bz2 $input_dir/output/{dde,icrmsd,rmsd,tfd}.csv \
+            tmp.sqlite.bz2 $input_dir/output/{dde,rmsd,tfd}.csv \
             $input_dir/output/{dde,rmsd,rmsd_cdf,tfd,tfd_cdf,bonds,angles,dihedrals,impropers}.png \
             env.yaml main.py $input_files
 

main.py

@@ -25,10 +25,10 @@ def make_csvs(store, forcefield, out_dir):
     store.get_rmsd(forcefield, skip_check=True).to_csv(f"{out_dir}/rmsd.csv")
     print("getting TFDs")
     store.get_tfd(forcefield, skip_check=True).to_csv(f"{out_dir}/tfd.csv")
-    print("getting internal coordinate RMSDs")
-    store.get_internal_coordinate_rmsd(forcefield, skip_check=True).to_csv(
-        f"{out_dir}/icrmsd.csv"
-    )
+    # print("getting internal coordinate RMSDs")
+    # store.get_internal_coordinate_rmsd(forcefield, skip_check=True).to_csv(
+    #     f"{out_dir}/icrmsd.csv"
+    # )
 
 
 def _main(forcefield, dataset, sqlite_file, out_dir, procs, invalidate_cache):

plot.py

@@ -30,9 +30,10 @@ def load_bench(d: Path) -> pandas.DataFrame:
     rmsd.columns = ["rec_id", "rmsd"]
     tfd = pandas.read_csv(d / "output" / "tfd.csv")
     tfd.columns = ["rec_id", "tfd"]
-    icrmsd = pandas.read_csv(d / "output" / "icrmsd.csv")
-    icrmsd.columns = ["rec_id", "bonds", "angles", "dihedrals", "impropers"]
-    ret = dde.merge(rmsd).pipe(DF.merge, tfd).pipe(DF.merge, icrmsd)
+    # icrmsd = pandas.read_csv(d / "output" / "icrmsd.csv")
+    # icrmsd.columns = ["rec_id", "bonds", "angles", "dihedrals", "impropers"]
+    # ret = dde.merge(rmsd).pipe(DF.merge, tfd).pipe(DF.merge, icrmsd)
+    ret = dde.merge(rmsd).pipe(DF.merge, tfd)
     print(f"loaded {ret.shape} rows for {d}")
     return ret
 
@@ -148,7 +149,7 @@ def plot(ffs, out_dir: str):
     plot_ddes(dfs, names, out_dir)
     plot_rmsds(dfs, names, out_dir)
     plot_tfds(dfs, names, out_dir)
-    plot_icrmsds(dfs, names, out_dir)
+    # plot_icrmsds(dfs, names, out_dir)
 
 
 @click.command()

submissions/2025-02-04-Sage-2.2.1/input.yaml

@@ -0,0 +1,3 @@
+forcefield: openff_unconstrained-2.2.1.offxml
+datasets:
+    - datasets/OpenFF-Lipid-Optimization-Training-Supplement-v1.0/cache.json