Ctrl+K

Quickstart

Contents

Quickstart#

To get started with Chemprop, first install the package using the instructions in the Installation section. Once you have Chemprop installed, you can train a model on your own data or use the pre-packaged solubility dataset to get a feel for how the package works.

Let’s use the solubility data that comes pre-packaged in the Chemprop directory:

$ head tests/data/regression.csv
smiles,logSolubility
OCC3OC(OCC2OC(OC(C#N)c1ccccc1)C(O)C(O)C2O)C(O)C(O)C3O,-0.77
Cc1occc1C(=O)Nc2ccccc2,-3.3
CC(C)=CCCC(C)=CC(=O),-2.06
c1ccc2c(c1)ccc3c2ccc4c5ccccc5ccc43,-7.87
c1ccsc1,-1.33
c2ccc1scnc1c2,-1.5
Clc1cc(Cl)c(c(Cl)c1)c2c(Cl)cccc2Cl,-7.32
CC12CCC3C(CCc4cc(O)ccc34)C2CCC1O,-5.03
ClC4=C(Cl)C5(Cl)C3C1CC(C2OC12)C3C4(Cl)C5(Cl)Cl,-6.29
...

Now we’re ready to train a simple Chemprop model:

chemprop train --data-path tests/data/regression.csv \
    --task-type regression \
    --output-dir train_example

This will train a model on the solubility dataset (tests/data/regression.csv) and save the model and training logs in the train_example directory. You should see some output printed to your terminal that shows the model architecture, number of parameters, and a progress bar for each epoch of training. At the end, you should see something like:

───────────────────────────────────────────────────────
   Test metric             DataLoader 0
───────────────────────────────────────────────────────
    test/mse             0.7716904154601469
───────────────────────────────────────────────────────

With our trained model in hand, we can now use it to predict solubilities of new molecules. In the absence of additional data, for demonstration purposes, let’s just test on the same molecules that we trained on:

chemprop predict --test-path tests/data/regression.csv \
    --model-path train_example/model_0/best.pt \
    --preds-path train_example/predictions.csv

This should output a file train_example/predictions_0.csv containing the predicted log(solubility) values for the molecules contained in tests/data/regression.csv.

$ head train_example/predictions_0.csv
smiles,logSolubility,pred_0
OCC3OC(OCC2OC(OC(C#N)c1ccccc1)C(O)C(O)C2O)C(O)C(O)C3O,-0.77,-1.0349703
Cc1occc1C(=O)Nc2ccccc2,-3.3,-3.0304263
CC(C)=CCCC(C)=CC(=O),-2.06,-2.0320206
...

Given that our test data is identical to our training data, it makes sense that the predictions are similar to the ground truth values.

In the rest of this documentation, we’ll go into more detail about how to:

Summary#

  • Install Chemprop using the instructions in the Installation section

  • Train a model with chemprop train --data-path <input_path> --task-type <task> --output-dir <dir>

  • Use a saved model for prediction with chemprop predict --test-path <test_path> --checkpoint-dir <dir> --preds-path <path>

Contents