# ForMileS: A Python Open-Source Program to Generate Molecular Structures for Tandem Mass Spectrometry Fragment Ions

## TL;DR

ForMileS is an open-source Python tool that generates molecular structures for fragment ions in tandem mass spectrometry, helping chemists better understand fragmentation processes.

## Contribution

ForMileS introduces a streamlined workflow with a simplified branch-and-bound algorithm for generating fragment ion structures with specific constraints.

## Key findings

- ForMileS successfully generates plausible structures for fragments like those of Polypropylene Glycol Octamer and dipropylene glycol dimethyl ether.
- Linear double-bonded structures are energetically favored over cyclic ones in DGDE fragments.
- The tool's performance is limited by exhaustive combinatorial charge generation and needs optimization for larger molecules.

## Abstract

Tandem mass spectrometry
is a central analytical tool
in chemistry,
yet the fragmentation mechanisms underlying collision-induced dissociation
remain incompletely understood. A key challenge is predicting fragment
ion structures while preserving the essential structural features
of the precursor ion. This paper introduces ForMileS (Formation of
Mass SMILES), a streamlined Python open-source workflow for generating
fragment ion structures with precursor-specific constraints from tandem
mass spectrometry data. ForMileS employs a simplified branch-and-bound
algorithm, accepting molecular formula, charge state, exact mass,
and a base scaffold in SMILES format as input, along with parameters
for branching, cyclicity, and bond types, via a graphical user interface.
We demonstrate its application to the three main fragments of Polypropylene
Glycol Octamer (PPG8), discussing the critical role of the base molecular
scaffold (BMS) in the final structure set. Relative energy calculations
using Density Functional Theory confirm the presence of expected structures,
highlighting the lowest energy conformers. When applied to the smallest
fragment of dipropylene glycol dimethyl ether (DGDE), ForMileS reveals
that only linear double-bonded or cyclic structures are plausible,
with the former being energetically favored. While successfully generating
plausible structures, the exhaustive combinatorial charge generation
step and the unrefined branch-and-bound method limit ForMileS’s
performance, restricting its applicability to small molecules like
C6O3H19. This highlights the importance
of future performance optimization through heuristics and energetic
filters.