Paper / PDF
TL;DR
In this paper we present:
- weighted mass correlation networks (WMCN) identified for inclusions in the bulk of the analyte material
- sliding window centrality measure and a divergence of the modularity score on spots with more that one chemical entities present
- kernel density estimation on ion-intensity regions for specific compounds
- Van-Krevelen metric for organic vs. inorganic materials
- PCA scores and loadings for chemical compounds identified in the sample
- We performed a competitive test of 25 supervised machine learning models to achieve a 99% accuracy rates for identification of investigated materials (bio-organic Precambrian kerogen vs. inorganic host mineral)