Proteins play a diverse and crucial role in essential physiological pro-
cesses, intricately interacting with other proteins and biomolecules such
as lipids or ligands. Understanding the structures and mechanisms of
these biomolecular systems is crucial to understanding their specific
functions. Experimental techniques are constantly improving. It is im-
perative to simultaneously develop computational methods to bridge
the gap between raw data and meaningful results. The main focus of
this thesis is on the development of analytical models and integrative
computational tools to fully exploit the wealth of structural information
that can be extracted from small-angle scattering (SAS) data. The thesis
also explores how molecular dynamics (MD) simulations of proteins can
enhance the interpretation of experimental data by providing insights
which are not accessible through experiments alone.
Firstly, the thesis focuses on the advancement of size-exclusion chro-
matography coupled with small-angle x-ray scattering (SEC-SAXS) and
introduces a novel procedure to investigate underlying structural dis-
tributions within a single species. It shows how an analytical model
can be refined against many frames from the same SEC-SAXS data sets
simultaneously to provide more robust fit results. The procedure is
applied to study populations of nanodiscs. This thesis also explores dif-
ferent methods for modelling flexible membrane proteins embedded in
nanodiscs. Flexible particles pose a challenge for SAS analysis, since the
scattering signal is averaged over an ensemble of conformations. There-
fore, an advanced semi-analytical model accounting for conformational
diversity was built for the human growth hormone receptor (GHR) in a
nanodisc and refined against SAXS data. The thesis goes on to discuss
methods for ensemble modelling of membrane proteins in nanodiscs.
In the case of the GHR, a simulated ensemble of protein structures was
placed in an analytical nanodisc model with pre-determined parame-
ters. The averaged theoretical scattering from the ensemble was in good
agreement with the SAXS data. It is then shown how a novel method
based on point-cloud models and Fast Debye Sums can be used to refine
nanodisc parameters for an entire ensemble of protein structures in a
more accurate and computationally efficient manner.
Furthermore, this thesis delves into a comprehensive SAS study on the
interaction of α-Synuclein and negatively charged lipid structures. The
data suggest that the amphipathic properties of the protein can induce a
break down of the lipid structures into smaller disc- or rod-like particles.


Detailed model-free analysis as well analytical models were used to
characterise the structural transformations.
Finally, the focus is shifted away from SAS towards all atom simulations
of amyloid fibrils. Experimental Φ-values were employed to guide the
simulations to sample the transition state of amyloid fibril elongation.
The crucial interactions sites between the incoming monomer and fibril
end were identified to help shed light on the mechanisms of fibril
formation.