Home

Welcome to the ChemistryFeaturization.jl wiki!

This package is in development as part of the ACED project, funded by ARPA-E DIFFERENTIATE and coordinated by Carnegie Mellon University, in collaboration with Julia Computing, Citrine Informatics, and MIT.

ChemistryFeaturization.jl is meant to be a unified interface for translating atomic structures (molecules, crystals, etc.) into data structures and sets of features to be used in various machine learning models, such as AtomicGraphNets.jl.

It is intended as a "helper" package for these other packages that actually build the models, and NOT as a standalone tool (e.g. it does not itself implement any models). Hopefully, as we work to develop more pieces of DeepChem in Julia, it will facilitate testing different models on the same input data without extensive and tedious labor to build features compatible with different architectures...