STUDY OF ATYPICAL STRUCTURES ON PROTEIN ENERGY LANDSCAPE VISUALIZATION

The protein folding process is a well-known problem concerning how a protein reaches its native and functional state. Using Energy Landscape Visualization Method (ELViM) we are able to obtain a 2D-projection of the multidimensional conformational phase-space of a molecular dynamics (MD) trajectory. However, in the projection, we often have the appearance of atypical structures (outliers) that pull away its projection’s neighbors structures due to iterations of the dimensional reduction method (force scheme). To a better understanding and use of the method, it is essential to characterize these structures, explain their appearance, and if necessary, how to avoid them. The main objective of this work is to study the atypical structures of energy surfaces of the protein A using the ELViM, to explain the reason for their appearance and how to diminish their peculiar aspect on the projection. In this work, we analyzed the projection of the conformational phase-space of protein-A obtained with a Cα structure-based model MD simulations. The dynamics were performed at the protein folding temperature and analyzed using the ELViM. The results allowed the visualization of the energy landscape and the identification of its folding route. However, in a two-dimensional representation, some configurations often behave as outliers; that is, they are radially distant from all neighboring points such as "islands" that distort the overall landscape representation. The results showed us that the formation of these outliers appears when this structure is chosen later in the iterative distance calculations on the force scheme algorithm. The size of the island is also related to the region of the formation of then. After the outlier’s characterization, two different methods were proposed to diminish the atypical structures of the projection: i) to choose these structures first in the distance calculations, and ii) to not choose them for the force scheme iterations. The analysis of the structures of the island locations revealed that they do not have any singular characteristics, and it only occurs due to the multidimensional reduction method, with both proposed methods being able to reduce the number of atypical structures, allowing us to smooth the projection without losing the characteristic of the landscape.