A Paradigm Shift in Computational Modelling: From Managing Workflows to Making Discoveries

In my work as a computational chemist, I have spent years studying materials and building scientific software. I have seen how much time researchers can spend preparing calculations instead of exploring scientific questions.

Exploring a new material can involve constructing a structure, modifying its chemistry, preparing input files, moving between software packages, setting up calculations and analysing the results. Depending on the problem, preparing this workflow can take hours or days.

Scientific calculations take time. But I often ask how much of the surrounding preparation is necessary.

That question led me to found StructSub Ltd, a scientific software and AI company focused on materials discovery.

I first set out to build a tool for modifying molecular and materials structures without writing complex scripts. Replacing a building unit in a metal-organic framework, functionalising a molecule or substituting atoms in a crystal should not take hours of preparation.

But as I continued developing the platform, I realised that structural modification was only one part of a much larger problem.

Building a structure is only a first step. Researchers also need to understand its stability, interactions and useful properties. They need to explore possible materials before deciding what is worth testing in the lab.

StructSub brings structure generation, modification, simulation and analysis into one browser-based scientific workspace. Researchers can work with molecules, polymers, peptides, crystals, zeolites, metal-organic frameworks and covalent organic frameworks. They can build and modify structures, optimise geometries and investigate material properties.

The platform also supports calculations relevant to adsorption, gas separation, ionic transport, mechanical properties and other materials applications. By combining established computational methods with machine learning and AI, our goal is to make these capabilities easier to access without removing the scientific choices that researchers need to make.

Time matters to me. Better models and more powerful computers will help materials discovery, but so will making computational tools easier to use.

A researcher should be able to start with a scientific question, explore a structural idea, perform a calculation and interpret the results without spending most of their time managing files and software. What previously required several separate steps should, where possible, happen within one connected workflow.

This matters especially for researchers with strong scientific ideas but limited programming experience or access to specialist computing.

Scientific ideas should not be limited by the complexity of the tools needed to explore them.

I want StructSub to grow into a scientific workspace where researchers can describe a question, explore possible structures, run calculations and examine results in one place. AI could help interpret results and suggest what to investigate next, while researchers remain responsible for the scientific decisions.

StructSub is at an early stage. There is more to develop and validate. Reliable scientific software needs careful testing, transparent methods and clear limits on what computational predictions can tell us.

I founded StructSub to give researchers more time to test scientific ideas and less time managing disconnected tools.