Understanding and exploring macromolecular dynamics is crucial for drug development, studying biological functions, and protein engineering, and thus remain a core focus in structural biology. However, revealing partially-occupied states in X-ray crystallography faces challenges due to inherent averaging over a population of proteins in different states, obscuring low-occupancy states in experimentally determined electron densities. One way forward is offered by analysing multiple datasets together, in particular the variations of occupancies found between equivalent datasets, i.e. data collected from different crystals grown in equivalent experimental conditions. In this thesis, I consider real datasets as realisations of a distribution of possible occupancy sets, and try to fit a model against this data using maximum likelihood methods. The resulting network outputs maps resembling electron density maps, containing small differences but does not yet reveal underlying states for the residues. This is believed to be caused by a local minimum, as the model optimises the maps to replicate an average of the real data.
Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:liu-205216 |
Date | January 2024 |
Creators | Strid Holmertz, Ylva |
Publisher | Linköpings universitet, Institutionen för fysik, kemi och biologi |
Source Sets | DiVA Archive at Upsalla University |
Language | English |
Detected Language | English |
Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
Format | application/pdf |
Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0019 seconds