Peak selection matters in principal component analysis: A case study of syntrophic microbes

Abstract

In situ liquid time-of-flight secondary ion mass spectrometry (ToF-SIMS) is a powerful technique to study surface characterization of living biofilms in hydrated conditions. However, ToF-SIMS data analysis is still a great challenge in complicated bacterial biofilms, because many interference peaks from the medium may result in inaccurate interpretation. In this study, two syntrophic Geobacter populations are investigated using in situ liquid ToF-SIMS to reveal the biofilm surface changes between them due to direct interspecies electron transfer. By comparing spectral principal component analysis (PCA) results of all peaks and selected peaks, the authors find that spectral peak overlay is an effective strategy to reduce the matrix effect in handling complex ToF-SIMS data. Additionally, the spectral PCA results of high intensity and high resolution data obtained from liquid ToF-SIMS are compared. Selected peaks, amino acid peaks, and water cluster peaks spectral PCA produce nice separation among samples in both high intensity and high resolution data sets. However, the high resolution data show better separation between coculture planktonic and coculture aggregates, confirming that the higher mass accuracy is useful in the analysis of microbial samples. In conclusion, the results show that peak selection is critical for acquiring effective microbial information and interpretation of syntrophic Geobacter using spectral data from in situ liquid ToF-SIMS.