Update README.md

README.md

@@ -108,7 +108,7 @@ First, from the command line, install `MolFlux` library with `catboost` and `rdk
 
     pip install 'molflux[catboost,rdkit]'
 
-then load, featurize, split, fit, and evaluate the a catboost model
+then load, featurize, split, fit, and evaluate the catboost model
 
     import json
     from datasets import load_dataset
@@ -118,104 +118,7 @@ then load, featurize, split, fit, and evaluate the a catboost model
     from molflux.modelzoo import load_from_dict as load_model_from_dict
     from molflux.metrics import load_suite
 
-To sanitize molecules, import some modules
-
-    import pandas as pd
-    import numpy as np
-    import tqdm
-    import rdkit
-    from rdkit import Chem
-
-Correct SMILES strings for molecules that are causing SMILES parsing errors.
-
-    Long2023 = HematoxLong2023['train'].to_pandas()
-
-    Long2023 = Long2023.rename(columns={'canonical SMILES': 'SMILES'})
-
-    Long2023.loc[Long2023['SMILES'] == 'Sc1[nH]cnc-2[n+H]cnc1-2', 'SMILES'] = "SC1NCNC2NCNC21"
-    Long2023.loc[Long2023['SMILES'] == 'Sc1nc(N)nc2[n+H]c[nH]c12', 'SMILES'] = "C1=NC2=C(N1)C(=S)[NH+]=C(N2)N"
-    Long2023.loc[Long2023['SMILES'] == 'O(C1C(O)C(OC2C(O)C([N+H2]C)[C@](O)(C)CO2)C(N)CC1N)C1C([N+H3])CCC(C([N+H2]C)C)O1', 'SMILES'] = \
-     "CNC(C)[C@@H]1CC[C@@H](N)[C@H](O1)O[C@@H]1[C@@H](N)C[C@@H](N)[C@H](O[C@H]2OC[C@](C)(O)[C@H](NC)[C@H]2O)[C@H]1O"
-    Long2023.loc[Long2023['SMILES'] == 'O(C[C@@H](O)C[N+H2]C(C)C)c1c2c(ccc1)cccc2', 'SMILES'] = \
-     "O[C@@H](C[NH2+]C(C)C)COc2cccc1ccccc12"
-    Long2023.loc[Long2023['SMILES'] == 'OC1CC[N+H](CCC[N+@H]2c3c(Sc4c2cccc4)ccc(C#N)c3)CC1', 'SMILES'] = \
-     "c1ccc2c(c1)N(c3cc(ccc3S2)C#N)CCCN4CCC(CC4)O"
-    Long2023.loc[Long2023['SMILES'] == 'O=C(C)[C@@]1(O)CC(OC2OC(C)C(O)C([N+H3])C2)c2c(O)c3C(=O)c4c(C(=O)c3c(O)c2C1)cccc4', 'SMILES'] = \
-     "C[C@H]1[C@H]([C@H](C[C@@H](O1)O[C@H]2C[C@@](Cc3c2c(c4c(c3O)C(=O)c5ccccc5C4=O)O)(C(=O)C)O)N)O"
-    Long2023.loc[Long2023['SMILES'] == 'O(C)c1cc(NCCC[C@@H]([N+H3])C)c2ncccc2c1', 'SMILES'] = "C[C@@H](CCCNc1cc(cc2c1nccc2)OC)[NH3+]"
-    Long2023.loc[Long2023['SMILES'] == 'Fc1cc2/C(=C/c3c(C)c(C(=O)NCC[N+H](CC)CC)c(C)[nH]3)/C(=O)Nc2cc1', 'SMILES'] = \
-     "CC[NH+](CC)CCNC(=O)C1=C(NC(=C1C)C=C2C3=C(C=CC(=C3)F)NC2=O)C"
-    Long2023.loc[Long2023['SMILES'] == 'O=C([N+H2]c1cc(Nc2nc(-c3cnccc3)ccn2)c(C)cc1)c1ccc(CN2CCN(C)CC2)cc1', 'SMILES'] = \
-     "Cc1ccc(cc1Nc2nccc(n2)c3cccnc3)NC(=O)c4ccc(cc4)CN5CC[NH+](CC5)C"
-    Long2023.loc[Long2023['SMILES'] == 'Clc1cc2[n+H]c3c(c(N[C@H](CCC[N+H](CC)CC)C)c2cc1)cc(OC)cc3', 'SMILES'] = \
-     "Clc3ccc2c(c1cc(OC)ccc1[nH+]c2c3)N[C@@H](C)CCC[NH+](CC)CC"
-    Long2023.loc[Long2023['SMILES'] == 'O(CC[N+H2]C[C@H](O)COc1c-2c([N+H2]c3c-2cccc3)ccc1)c1c(OC)cccc1', 'SMILES'] = \
-     "COc1ccccc1OCCNCC(COc2cccc3c2c4ccccc4[nH]3)O"
-    Long2023.loc[Long2023['SMILES'] == 'O(C[C@@H]([N+H3])C)c1c(C)cccc1C', 'SMILES'] = \
-     "O(c1c(cccc1C)C)C[C@@H]([NH3+])C"
-    Long2023.loc[Long2023['SMILES'] == '[N+@H](CCCCC(C#N)(c1ccccc1)c1ccccc1)(CC[N+]1(C)CCOCC1)C', 'SMILES'] = \
-     "CN(CCCCC(C#N)(c1ccccc1)c2ccccc2)CC[N+]3(CCOCC3)C"
-    Long2023.loc[Long2023['SMILES'] == 'O=C(/N=C1/ON=[N+](N2[C@@H](C)CCC[C@H]2C)[C-H]/1)c1ccc(OC)cc1', 'SMILES'] = \
-     "C[C@@H]1CCC[C@@H](N1[n+]2[cH-]/c(=N\C(=O)c3ccc(cc3)OC)/on2)C"
-    Long2023.loc[Long2023['SMILES'] == 'O(C)c1cc2c([C@@H](O)[C@@H]3[N+@H]4[C@@H](C=C)[C@H](C3)CC4)ccnc2cc1', 'SMILES'] = \
-     "COC1=CC2=C(C=CN=C2C=C1)C(C3CC4CCN3C4C=C)O"
-
-Correct SMILES strings for molecules that are causing kekulization errors
-
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)C2=C3C(=CC=CC3=CC=C2)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=CC=CC3=CC=CC(=C23)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)NCCCl)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)NCCCl)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC=O)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC=O)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)CCCCl)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)CCCCl)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)C(Cl)(Cl)Cl)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)C(Cl)(Cl)Cl)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc([N+](=O)[O-])cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC([N+](=O)[O-])=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(N)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(N)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)Nc4ccc(Cl)cc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)NC4=CC=C(Cl)C=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)Nc4ccc(C#N)cc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)NC4=CC=C(C#N)C=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=S)Nc4ccc(Cl)cc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=S)NC4=CC=C(Cl)C=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(/N=C/c4ccc(C#N)cc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(/N=C/C4=CC=C(C#N)C=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=S)Nc4ccc(C#N)cc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=S)NC4=CC=C(C#N)C=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NCc4ccc5c(c4)OCO5)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NCC4=CC=C5OCOC5=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)NC(=O)c4ccccc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)NC(=O)C4=CC=CC=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'C[C@H]1Oc2cc(cnc2N)-c2c(nn(C)c2C#N)CN(C)C(=O)c2ccc(F)cc21', 'SMILES'] = 'C[C@H]1OC2=C(N)N=CC(=C2)C2=C(C#N)N(C)N=C2CN(C)C(=O)C2=CC=C(F)C=C21'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)Nc4cc5c6c(cccc6c4)C(=O)N(CCN(C)C)C5=O)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)NC4=CC5=C6C(=CC=CC6=C4)C(=O)N(CCN(C)C)C5=O)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NCc4ccc5c(c4)OCCO5)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NCC4=CC=C5OCCOC5=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)c4ccccc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)C4=CC=CC=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)NC(=O)CCl)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)NC(=O)CCl)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)C(F)(F)F)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)C(F)(F)F)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)Nc4ccc(OC(F)(F)F)cc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)NC4=CC=C(OC(F)(F)F)C=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(/N=C/c4cc(O)ccc4O)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(/N=C/C4=CC(O)=CC=C4O)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=S)Nc4ccc(OC(F)(F)F)cc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=S)NC4=CC=C(OC(F)(F)F)C=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NCc4cc5c(cc4[N+](=O)[O-])OCO5)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NCC4=CC5=C(C=C4[N+](=O)[O-])OCO5)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)Cc4ccc(Cl)cc4)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)CC4=CC=C(Cl)C=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(/N=C/c4cc5c(cc4[N+](=O)[O-])OCO5)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(/N=C/C4=CC5=C(C=C4[N+](=O)[O-])OCO5)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(/N=C/c4ccc5c(c4)OCO5)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(/N=C/C4=CC=C5OCOC5=C4)=CC3=CC=C2)C1=O'
-    Long2023.loc[Long2023['SMILES'] == 'CN(C)CCN1C(=O)c2cccc3cc(NC(=O)Nc4ccc5c(c4)OCO5)cc(c23)C1=O', 'SMILES'] = 'CN(C)CCN1C(=O)C2=C3C(=CC(NC(=O)NC4=CC=C5OCOC5=C4)=CC3=CC=C2)C1=O'
-
-Sanitize molecule with MolVS
-
-    $ pip install molvs
-
-    import molvs
-
-    standardizer = molvs.Standardizer()
-    fragment_remover = molvs.fragment.FragmentRemover()
-
-    Long2023['ID'] = [f"Long2023_{i}" for i in range(len(Long2023.index))]
-    
-    Long2023['X'] = [ \
-    rdkit.Chem.MolToSmiles(
-        fragment_remover.remove(
-        standardizer.standardize(
-        rdkit.Chem.MolFromSmiles(
-        smiles))))
-    for smiles in Long2023['SMILES']]
-
-most are because it includes the salt form and/or it is not neutralized
-
-    for id, alert in problems:
-        print(f"ID: {id}, problem: {alert[0]}")
-
-Split and evaluate the a catboost model
+Split and evaluate the catboost model
     
     split_dataset = load_dataset('maomlab/HematoxLong2023', name = 'HematoxLong2023')