msibi: Multistate Iterative Boltzmann Inversion

Iterative Boltzmann inversion (IBI) is a well-established and widely used method for deriving coarse-grained (CG) force fields that recreate the structural distributions of an underlying atomistic model. Multiple state IBI (MS-IBI), as introduced by Moore et al. (2014), addresses state-point transferability limitations of IBI by including distributions from multiple state points to inform the derived CG force field. Here, we introduce msibi, a pure Python package that implements the MS-IBI method for creating CG force fields for both intramolecular and intermolecular interactions. The package offers a user-friendly, Python-native API, eliminating the need for bash scripting and manual editing of input files. msibi is ultimately simulation engine agnostic, but uses the HOOMD-Blue simulation engine (Anderson et al., 2020) under-the-hood to perform iterative CG model simulations. This allows msibi to utilize graphical processing unit (GPU) acceleration without requiring users to manually compile GPU compatible code.