Similarity searching in macromolecular electron density maps

Identifying similarities and differences in images was and remains a major driving force of discovery in biological sciences. This thesis addresses said topic and attempts to leverage digital image analysis methods for similarity searching in macromolecular electron density maps, which can be acquired by X-ray crystallography and single particle cryo-electron microscopy. The work introduces a method coined siseek, which is based on the scale-invariant feature transform. Comprehensive experiments show that siseek registers intermediate and high resolution maps successfully requiring less time than other approaches. Additionally, siseek is used in proof of concept experiments for molecule recognition. The results demonstrate that siseek facilitates similarity searching in macromolecular electron density maps and show that the method can be used to aid the process of interpreting these maps.