Discerning protein similarities and folding dynamics through methods in statistical physics and molecular dynamics simulations

One of the most important tasks in biophysics is to understand proteins. It has long been observed that sequence similarity, structural similarity, and functional similarity among proteins are highly correlated. Inspired by this observation, sequence comparison is often employed to retrieve similar/related protein sequences. However, the central issue of assigning statistical significance to the proteins retrieved remains a challenging problem. How a directed polymer/percolation model can shed light in understanding the statistics of global sequence alignments, which is also the fundamental building block for multiple sequence alignment in many multiple alignment applications, is revealed in my thesis. In terms of understanding proteins' specific functionality, it is essential to know how protein sequences determine their unique three dimensional structures and folding kinetics. A careful analysis of the amino acid arrangements in proteins with known structures may provide insights to this issue. In my thesis, I will present a statistical analysis on tertiary contacts to gain more accurate estimates of the preference of amino acid interactions. This analysis reveals an unusually large contact between cysteines, indicating an effective attractive potential among them. The nontrivial role of cysteine-cysteine interactions in protein folding is discussed in my thesis. A new concept termed target-focusing is also introduced.