No matter if it’s the pharmaceutical industry, specialty chemicals or the perfume industry – they all face the same challenge every day: the identification of (unknown) substances. The related question “Which molecules are in my sample?” spans all areas of chemistry with the problem of compound identification being ubiquitous.
Does the sample really only contain the desired substances? And if not, are any impurities present admissible? Complex devices such as gas chromatography-mass spectrometers (GC-MS) are used to analyze these samples. However, these machines do not yield desired molecular structures directly, but complex spectrometric data. Since this data is highly convoluted – thereby challenging to be interpreted – today only one strategy exists for structure elucidation: Manual interpretation by trained chemistry experts. Even for them, these analyses take hours or days and due to rising personnel cost and scarcity they come with a price tag between €1,000 and €10,000 – for just a single sample.

To tackle this challenge, ChemInnovation develops chemically inspired AI models and specific cheminformatics algorithms. The team convolutes this basis into smart AI-Software for analytical chemical applications with the focus being on the evaluation of spectrometric data, and in particular on automating structure elucidation. Their first product, which was developed at the University of Münster as part of an EXIST research transfer, is METIS (Molecules from ElecTron Ionization Spectra). With the software, data can be analyzed fully automatically in <5 min.
This speeds up the evaluation process by a factor of x5 to x20, while 50% to 200% more molecular structures can be elucidated. While yielding the same or a higher quality of results, its use requires significantly less expertise to be used, especially compared to manual competitor software. The underlying algorithms have been developed in such a way that the software works reliably in a particularly wide range of applications, i.e. regardless of the exact questions and underlying chemistry.