PhD thesis submitted in fulfilment of the requirements for the degree of Doctor in Bio-engineering Sciences.
In a slightly melancholic mood (triggered by Google's DeepMind making significant strides in solving protein folding), I decided to publish my doctoral thesis — originally published on March 4th 2011 — to Medium (on April 8th 2019). I spent 5 odd years analysing and predicting protein structures, in particular in interactions with small amino acid based compounds called peptides. My thesis shares and summarises insights and showed then state-of-the-art predictive capabilities on protein-peptide interactions, many of which are already or will soon be eclipsed by advances in the field.
On the cover: The image shows the peptide binding site of a PDZ domain, superposed with similar arrangements of polypeptide fragments taken from monomeric proteins, demonstrating that protein-peptide interactions follow the same architectural patterns as those governing monomeric proteins. The puzzle in the background aims to accentuate the generality of this finding. Fragments superposed on the peptide are shown in green, fragments covering the PDZ binding site are shown in red (see Vanhee et al., 1128). Design of the cover by Antonio De Marco. Molecular graphics by Peter Vanhee using Yasara and PovRay.
Whilst I have long left the field of bioengineering sciences, and in particular computational biology, a long time ago, I believe I am still carrying many of the hard-learnt lessons of completing a PhD into my professional career as technologist.
Thesis Summary
Proteins are by far the most versatile and complex molecules in the cell. It is commonly accepted that protein function directly relates to three-dimensional structure, which in turn is dependent on the specific amino acid sequence of the protein. Peptides are short sequences of amino acids that perform a myriad of functions and are estimated to be involved in up to 40% of all protein-protein interactions. The lack of structural evidence for many of these peptide interactions however has hindered the functional annotation of this important class of molecules and the development of peptides as therapeutics. In this thesis, we propose the use of small, recurrent polypeptide fragments as one way of solving the lack of protein-peptide structures. We show that protein-peptide binding sites can be modelled at high resolutions using fragment interactions and provide two methods for the de novo prediction of protein loops and peptide structure. The developments presented in this work provide a valuable alternative to experimental high-resolution structure elucidation of target protein-peptide complexes, bringing closer the possibility of in-silico designed peptides for therapeutic applications.
View the full thesis on SlideShare.