Non-negative matrix factorisation of large mass spectrometry datasets

Abstract

The development of state-of-art time-of-flight secondary ion mass spectrometry (ToF-SIMS) results in extremely large datasets. In order to perform multivariate analysis of such datasets without loss of mass and spatial resolution, appropriate data handling methods must be developed. The work in this paper presents an approach that can be taken to perform non-negative matrix factorisation (NMF) of large ToF-SIMS datasets. A large area stage raster scan of a chemically contaminated fingerprint is used as an example and the results show that the fingerprint signal was successfully separated from the substrate signal. Pre-processing challenges and artefacts that arises from the results are also discussed and an alternative approach, using the MapReduce programming model, is suggested for even larger datasets.