PEAS: An Application for Autonomous Precision Conformation Sampling

TL;DR

PEAS is a new Python application that automates and simplifies molecular modeling workflows for computational chemistry and biology.

Contribution

The novel contribution is the integration of multiple validated tools into a unified, user-friendly application for autonomous conformation sampling.

Findings

PEAS combines SEER, Confab, and CCS Focusing to streamline molecular modeling workflows.

The application reduces user intervention by integrating modeling engines vertically.

PEAS has been validated for efficiency and acceptable accuracy in pre-quantum mechanical processing.

Abstract

Molecular modeling tools are routinely utilized in computational
chemistry and computational biology projects. The ongoing advancements
in hardware and software have made modeling diverse chemical systems
more accurate and computationally affordable. However, with many software
tools available to perform multiple relevant tasks, selecting the
best workflow can become daunting in itself. In one of our recent
works, we developed a workflow to assign chemical structures to experimental
ion mobility mass spectrometry collisional cross-section (CCS) values.
This requires multiple steps, including protonation state assignment,
relevant conformational search, and conformation similarity filtering,
to deliver a manageable workload for downstream quantum mechanical
(QM) calculations. To simplify running our workflow, we have developed
an open-source, user-friendly Python application called PEAS (precise ensemble autonomous sampling) that effectively streamlines the result chain through
vertical modeling engine integration to limit user intervention. Since
the crucial steps prior to quantum mechanical processing in modeling
are charge state determination and relevant conformation sampling,
we have therefore incorporated SEER (charge state predictor), Confab
(conformation generator), and CCS Focusing (conformer filtering) into
the development of PEAS. These engines have been separately validated
and confirmed for efficiency and acceptable accuracy, and thus, we
report that their unified performance also delivers similar outcomes.
Documentation for PEAS and its Google Colab executable platform is
available at https://github.com/mitkeng/peas.

Figures