Biological materials typically show a large variation in local structure and local composition. As a versatile and information-rich method, this variation can be characterized using scanning X-ray scattering at synchrotron beamlines with microfocus beam. Typical applications include the simultaneous characterization of structure using small- and wide-angle scattering, and composition from X-ray fluorescence. For each spot, one thus obtains multimodal data including detector images plus fluorescence spectra, which for one experimental campaign can lead to data sizes of several TB. In several collaborations, we have collected data on different biological materials, including bone, teeth, tissues, or also plant seeds, which we now aim to consistently and safely curate, reduce and analyze with a common analysis framework. We plan in particularly to characterize the interplay of composition and local structure by employing tools from machine learning as well as correlation analysis. To this end, we will first implement a data pipeline to treat the 2d detector images in terms of orientation and hierarchical structure. The resulting parameter will consist of 2d spatial maps of the samples which characterize specific structural parameters. These will be compared to the compositional analysis from standard XRF analysis which results in 2d spatial maps of element concentration. Using matrix decomposition, we aim to detect typical patterns in these stack of 2d spatial maps. In addition, we will use autocorrelations to characterize structural domains. For all samples, these features will be compared between different treatment protocols of the samples to learn about aspects of health and food technology.