Contains scripts to find pairs of molecules with isim for siamese neural networks (SNN). Siamese neural networks receive as input pairs of objects. For large scale datasets traning the SNN on all the posible pairs is inefficient and resource consuming. Model performance has been found to not be as affected if less pairs are used to train the model.
In this module, we aim to use iSIM measurements to explore the chemical space of sets and generate posible pairs to train the models.