Datasets:

Modalities:
Text
Formats:
csv
ArXiv:
Libraries:
Datasets
Dask
License:
SMILES
stringlengths
15
57
SPLIT
stringclasses
1 value
CCCS(=O)c1ccc2[nH]c(=NC(=O)OC)[nH]c2c1
train
CC(C)(C)C(=O)C(Oc1ccc(Cl)cc1)n1ccnc1
train
Cc1c(Cl)cccc1Nc1ncccc1C(=O)OCC(O)CO
train
Cn1cnc2c1c(=O)n(CC(O)CO)c(=O)n2C
train
CC1Oc2ccc(Cl)cc2N(CC(O)CO)C1=O
train
CCOC(=O)c1cncn1C1CCCc2ccccc21
train
COc1ccccc1OC(=O)Oc1ccccc1OC
train
O=C1Nc2ccc(Cl)cc2C(c2ccccc2Cl)=NC1O
train
CN1C(=O)C(O)N=C(c2ccccc2Cl)c2cc(Cl)ccc21
train
CCC(=O)c1ccc(OCC(O)CO)c(OC)c1
train
Cc1nc2c([nH]1)c(=O)n(C)c(=O)n2CC1CC=CCC1
train
COc1cc2c(cc1O)N=CC1CCC(O)N1C2=O
train
COc1c(C)cnc(CS(=O)c2nc3ccccc3[nH]2)c1C
train
COc1cc(C)c(Cc2cnc(N)nc2N)cc1OC
train
O=C1Nc2ccc(Cl)cc2C(c2ccccc2)=NC1O
train
CC1CC(OC(=O)CN2CCCC2=O)CC(C)(C)C1
train
O=C(C1CCCCC1)N1CC(=O)N2CCc3ccccc3C2C1
train
COC(=O)c1c[nH]c2cc(OC(C)C)c(OC(C)C)cc2c1=O
train
CCC1NC(=O)c2cc(S(N)(=O)=O)c(Cl)cc2N1
train
CN1C(=O)C(O)N=C(c2ccccc2)c2cc(Cl)ccc21
train
CC(C)(C)C(=O)C(Oc1ccc(Cl)cc1)n1cncn1
train
COc1cc(C(=O)N2CCOCC2)cc(OC)c1OC
train
CC(c1ncncc1F)C(O)(Cn1cncn1)c1ccc(F)cc1F
train
COc1ccc(C=C2CCCN=C2c2cccnc2)c(OC)c1
train
CC(=O)Nc1ccc(S(=O)(=O)c2ccc(NC(C)=O)cc2)cc1
train
CC(=O)Oc1cc(C)c(OC(C)=O)c2ccccc12
train
CC(=O)Oc1cccc(OC(C)=O)c1OC(C)=O
train
CC1(C)C(=O)Nc2cc3nc(-c4ccncc4)[nH]c3cc21
train
CC(C)C=c1[nH]c(=O)c(=Cc2ccccc2)[nH]c1=O
train
NC(=O)Nc1ccc(S(=O)(=O)c2ccc(N)cc2)cc1
train
CC(C)(C#N)c1cc(Cn2cncn2)cc(C(C)(C)C#N)c1
train
CNC(=O)Oc1ccccc1C(=O)Nc1ccccc1
train
COc1ccc(-c2nnc(C)nc2-c2ccc(OC)cc2)cc1
train
Nc1ncnc2c1ncn2Cc1c(F)cccc1Cl
train
Clc1ccc(C2(Cn3cncn3)OCCO2)c(Cl)c1
train
Cc1cc(Cc2cnc(N)nc2N)c2cccnc2c1N(C)C
train
NS(=O)(=O)c1ccc(NCc2ccccc2)cc1
train
CC1(C)C=C(n2ccccc2=O)c2cc(C#N)ccc2O1
train
O=C1CN=C(c2ccccn2)c2cc(Br)ccc2N1
train
CN(CCO)c1nc2c(c(=O)n(C)c(=O)n2C)n1C
train
CCC1(O)C(=O)OCc2c1cc1n(c2=O)Cc2cc3ccccc3nc2-1
train
CN1C(=O)CC(=O)N(c2ccccc2)c2cc(Cl)ccc21
train
NS(=O)(=O)c1cc2c(cc1Cl)CN(C1CCCCC1)C2=O
train
CCc1cc2c(s1)N(C)C(=O)CN=C2c1ccccc1Cl
train
CC1(C)Oc2ccc(C#N)cc2C(N2CCCC2=O)C1O
train
Nc1nc2c(ncn2C2CC(CO)C2CO)c(=O)[nH]1
train
CC12CCC3=C(CCc4cc(O)ccc43)C1CCC2=O
train
CNC(=O)c1ccc(Cl)c(S(=O)(=O)NC)c1
train
NC(=O)c1cc(Br)cc(Br)c1O
train
O=C(c1ccccc1)c1cccc(NS(=O)(=O)C(F)F)c1
train
CCCN(CCC)S(=O)(=O)c1ccc(C)cc1
train
O=C1OCC(Cc2cccc(O)c2)C1Cc1cccc(O)c1
train
CCc1ccccc1-n1c(C)nc2ccccc2c1=O
train
CCc1cc2c(s1)-n1c(C)nnc1CN=C2c1ccccc1Cl
train
CCc1nc(N)nc(N)c1-c1ccc(Cl)c(Cl)c1
train
Cc1c(O)c(C)c2c(c1O)C(=O)CC(c1ccc(O)cc1)O2
train
CN1C(=O)CN=C(c2ccccc2F)c2cc(Cl)ccc21
train
Nc1nc2c(ncn2COC(CO)CO)c(=O)[nH]1
train
COc1c(OC)c(OC(C)=O)c2cc(Cl)ccc2c1OC(C)=O
train
Cc1nc2ccccc2c(=O)n1-c1ccccc1Cl
train
CCOc1ccc(-n2c(C)nc3ccccc3c2=O)cc1
train
Cc1nc2ccccc2c(=O)n1-c1ccc(Cl)cc1
train
Cc1cc2c(cc1S(N)(=O)=O)S(=O)(=O)CCC2
train
Cc1nc(N)nc(N)c1-c1ccc(Cl)c(Cl)c1
train
Cc1ccc(C)n1Nc1ccc(N2CCOCC2)nn1
train
O=C1CC(c2ccc(O)cc2)Oc2cc(O)cc(O)c21
train
Nc1ncnc2c1ncn2C1C=C(CO)C(O)C1O
train
O=C(c1ccccc1Nc1ccncc1)N1CCCCC1
train
CC1=NC(=O)C(C#N)=CC1=C1C=CN2C=CNC2=C1
train
CC(=O)Nc1ccc(OC(=O)c2cccs2)cc1
train
Nc1nc2c(ncn2CCC(CO)CO)c(=O)[nH]1
train
OCC1(CO)COC(C(Cl)(Cl)Cl)OC1
train
CCN(CC)C(=O)c1ccccc1C(=O)N(CC)CC
train
COc1ccc(C2Cc3cccc(O)c3C(=O)O2)cc1O
train
C#CCN1C(=O)CN=C(c2ccccc2)c2cc(Cl)ccc21
train
Cc1c2c(cn1C)NC(=O)CN=C2c1ccccc1
train
CCn1nc(C)c2c1C(c1ccccc1)=NCC(=O)N2
train
Cc1c(NC(=O)c2ccccc2O)c(=O)n(-c2ccccc2)n1C
train
CCn1ccc(NS(=O)(=O)c2ccc(N)cc2)nc1=O
train
Cc1cnc(NS(=O)(=O)c2ccc(N)cc2)nc1
train
Nc1ccc(S(=O)(=O)Nc2cnccn2)cc1
train
CC(=O)N(c1onc(C)c1C)S(=O)(=O)c1ccc(N)cc1
train
NS(=O)(=O)c1ccc(N2CCCCS2(=O)=O)cc1
train
CN1C(=O)CC(=O)N(c2ccccc2)c2cc(C(F)(F)F)ccc21
train
Cc1nn(C)c2c1C(c1ccccc1F)=NCC(=O)N2C
train
CS(=O)(=O)c1ccc(-c2cn3ccccc3n2)cc1
train
Cc1nn(C)c2c1C(c1cccc(Cl)c1)=NCCN2
train
NS(=O)(=O)c1cc(C(=O)c2ccc(O)cc2)cs1
train
COC1C(O)C(CO)OC1n1cnc2c(N)ncnc21
train
N#Cc1ccc(Nc2ncnc3c2CCC3O)cc1
train
CN(C)c1ncnc2c1ncn2Cc1ccccc1
train
CCOC(=O)n1nc(N)c2c1CN(C(=O)OC(C)(C)C)C2
train
O=C1c2cccnc2CN1Cc1c(F)cccc1F
train
COc1ccc(CN2Cc3ncccc3C2=O)cc1
train
O=C1c2cccnc2CN1Cc1ccc(Cl)cc1
train
O=C1c2cccnc2CN1Cc1cccc(Cl)c1
train
O=C1c2cccnc2CN1Cc1ccccc1C(F)(F)F
train
CCOc1ccnc(N2CCN(C(=O)c3sccc3C#N)CC2)n1
train
CCc1nc(C#N)c(N2CCc3ccccc3CC2)nc1C
train
COc1cccc(-c2csc(-c3cccnc3)n2)c1
train

Molecular Sets (MOSES): A benchmarking platform for molecular generation models

Deep generative models are rapidly becoming popular for the discovery of new molecules and materials. Such models learn on a large collection of molecular structures and produce novel compounds. In this work, we introduce Molecular Sets (MOSES), a benchmarking platform to support research on machine learning for drug discovery. MOSES implements several popular molecular generation models and provides a set of metrics to evaluate the quality and diversity of generated molecules. With MOSES, we aim to standardize the research on molecular generation and facilitate the sharing and comparison of new models.

For more details, please refer to the paper.

If you are using MOSES in your research paper, please cite us as

@article{10.3389/fphar.2020.565644,
  title={{M}olecular {S}ets ({MOSES}): {A} {B}enchmarking {P}latform for {M}olecular {G}eneration {M}odels},
  author={Polykovskiy, Daniil and Zhebrak, Alexander and Sanchez-Lengeling, Benjamin and Golovanov, Sergey and Tatanov, Oktai and Belyaev, Stanislav and Kurbanov, Rauf and Artamonov, Aleksey and Aladinskiy, Vladimir and Veselov, Mark and Kadurin, Artur and Johansson, Simon and  Chen, Hongming and Nikolenko, Sergey and Aspuru-Guzik, Alan and Zhavoronkov, Alex},
  journal={Frontiers in Pharmacology},
  year={2020}
}

Dataset

We propose a benchmarking dataset refined from the ZINC database.

The set is based on the ZINC Clean Leads collection. It contains 4,591,276 molecules in total, filtered by molecular weight in the range from 250 to 350 Daltons, a number of rotatable bonds not greater than 7, and XlogP less than or equal to 3.5. We removed molecules containing charged atoms or atoms besides C, N, S, O, F, Cl, Br, H or cycles longer than 8 atoms. The molecules were filtered via medicinal chemistry filters (MCFs) and PAINS filters.

The dataset contains 1,936,962 molecular structures. For experiments, we split the dataset into a training, test and scaffold test sets containing around 1.6M, 176k, and 176k molecules respectively. The scaffold test set contains unique Bemis-Murcko scaffolds that were not present in the training and test sets. We use this set to assess how well the model can generate previously unobserved scaffolds.

Downloads last month
55

Models trained or fine-tuned on katielink/moses