data

Specifications of Dataset Download in Geom3D

We provide both the raw and processed data at this HuggingFace link.

PCQM4Mv2

mkdir -p pcqm4mv2/raw
cd pcqm4mv2/raw
wget http://ogb-data.stanford.edu/data/lsc/pcqm4m-v2-train.sdf.tar.gz
tar -xf pcqm4m-v2-train.sdf.tar.gz


wget http://ogb-data.stanford.edu/data/lsc/pcqm4m-v2.zip
unzip pcqm4m-v2.zip
mv pcqm4m-v2/raw/data.csv.gz .
rm pcqm4m-v2.zip
rm -rf pcqm4m-v2 

GEOM

wget https://dataverse.harvard.edu/api/access/datafile/4327252
mv 4327252 rdkit_folder.tar.gz
tar -xvf rdkit_folder.tar.gz

Molecule3D

Install it following the google drive link here.

QM9

Automatically installed under folder .QM9/raw.

MD17

Automatically installed under folder ./MD17.

In March 2023 (or even earlier), they updated the MD17 FTP site, and the previous datasets are missing. We may need to keep and upload a version to the website.

rMD17

Download the dataset from this link, and put the file 12672038.zip under ./rMD17 folder.

• unzip 12672038.zip
• tar xjf rmd17.tar.bz2
• mv rmd17/npz_data .
• mv rmd17/splits .

COLL

We use this repo: git@github.com:TUM-DAML/gemnet_pytorch.git.

LBA/PDBBind

mkdir -p lba/raw
mkdir -p lba/processed
cd lba/raw
# wget http://www.pdbbind.org.cn/download/pdbbind_v2015_refined_set.tar.gz
# wget http://www.pdbbind.org.cn/download/pdbbind_v2018_refined.tar.gz
# wget http://www.pdbbind.org.cn/download/pdbbind_v2019_refined.tar.gz
# wget https://zenodo.org/record/4914718/files/LBA-split-by-sequence-identity-30-indices.tar.gz

wget http://www.pdbbind.org.cn/download/PDBbind_v2020_refined.tar.gz
tar -xzvf PDBbind_v2020_refined.tar.gz

wget https://zenodo.org/record/4914718/files/LBA-split-by-sequence-identity-30.tar.gz
tar -xzvf LBA-split-by-sequence-identity-30.tar.gz
mv split-by-sequence-identity-30/indices ../processed/
mv split-by-sequence-identity-30/targets ../processed/

LEP

mkdir -p lep/raw
mkdir -p lep/processed
cd lep/raw

wget https://zenodo.org/record/4914734/files/LEP-raw.tar.gz
tar -xzvf LEP-raw.tar.gz
wget https://zenodo.org/record/4914734/files/LEP-split-by-protein.tar.gz
tar -xzvf LEP-split-by-protein.tar.gz

MoleculeNet dataset

wget http://snap.stanford.edu/gnn-pretrain/data/chem_dataset.zip
unzip chem_dataset.zip
dataset_list=(tox21 toxcast clintox bbbp sider muv hiv bace)
for dataset in "${dataset_list[@]}"; do
    mkdir -p molecule_datasets/"$dataset"/raw
    cp dataset/"$dataset"/raw/* molecule_datasets/"$dataset"/raw/
done
rm -rf dataset

wget -O malaria-processed.csv https://raw.githubusercontent.com/HIPS/neural-fingerprint/master/data/2015-06-03-malaria/malaria-processed.csv
mkdir -p ./molecule_datasets/malaria/raw
mv malaria-processed.csv ./molecule_datasets/malaria/raw/malaria.csv

wget -O cep-processed.csv https://raw.githubusercontent.com/HIPS/neural-fingerprint/master/data/2015-06-02-cep-pce/cep-processed.csv
mkdir -p ./molecule_datasets/cep/raw
mv cep-processed.csv ./molecule_datasets/cep/raw/cep.csv

EC & FOLD

Check this link.

• ProtFunct is for task EC
• HomologyTAPE is for task FOLD

Or

• cd EC; python download.py
• cd FOLD; python download.py

MatBench

mkdir MatBench
cd MatBench

wget https://figshare.com/ndownloader/files/17494820
mv 17494820 expt_is_metal.json.gz
gzip -d expt_is_metal.json.gz

wget https://figshare.com/ndownloader/files/17494814
mv 17494814 expt_gap.json.gz
gzip -d expt_gap.json.gz

wget https://figshare.com/ndownloader/files/17494637
mv 17494637 glass.json.gz
gzip -d glass.json.gz

wget https://figshare.com/ndownloader/articles/9755486/versions/2
mv 2 perovskites.json.gz
unzip perovskites.json.gz
rm perovskites.json.gz
rm 17494805_perovskites.json.gz
gzip -d 17494808_perovskites.json.gz
mv 17494808_perovskites.json perovskites.json

wget https://figshare.com/ndownloader/files/17476067
mv 17476067 dielectric.json.gz
gzip -d dielectric.json.gz

wget https://figshare.com/ndownloader/files/17476064
mv 17476064 log_gvrh.json.gz
gzip -d log_gvrh.json.gz

wget https://figshare.com/ndownloader/files/17476061
mv 17476061 log_kvrh.json.gz
gzip -d log_kvrh.json.gz

wget https://figshare.com/ndownloader/files/17476046
mv 17476046 jdft2d.json.gz
gzip -d jdft2d.json.gz

wget https://figshare.com/ndownloader/files/17476040
mv 17476040 steels.json.gz
gzip -d steels.json.gz

wget https://figshare.com/ndownloader/files/17476037
mv 17476037 phonons.json.gz
gzip -d phonons.json.gz

wget https://figshare.com/ndownloader/files/17476034
mv 17476034 mp_is_metal.json.gz
gzip -d mp_is_metal.json.gz

wget https://figshare.com/ndownloader/files/17476028
mv 17476028 mp_e_form.json.gz
gzip -d mp_e_form.json.gz

wget https://figshare.com/ndownloader/files/17084741
mv 17084741 mp_gap.json.gz
gzip -d mp_gap.json.gz

The dataset size can match with MatBenchmark v0.1.

QMOF

mkdir QMOF
cd QMOF
wget https://figshare.com/ndownloader/articles/13147324/versions/13
mv 13 qmof_database_v13.zip
unzip qmof_database_v13.zip
unzip qmof_database.zip

cd qmof_database
python xyz_to_cifs.py
cd ../..

Or follow this link for prediction on QMOF DB v13.