04-11 March 2020
University of Roma 3, Rome, Italy
Dr. Ivan Coluzza
CIC BioMagune, San Sebastian (Donostia), Spain
Aim of the mission
Aim of the exchange is the finalization of the development of a novel coarse-graining strategy aimed at extracting protein-polymer interactions from the Ligand Binding site Database by means of statistical
mechanics tools. Through a Direct Coupling Analysis (DCA) method , the goal of the project is the development of a matrix of effective interactions obtained by the frequency matrices provided by Gabriele Macari.
 S. Cocco et al., Reports on Progress in Physics (2018)
Summary of the Results
The preliminary step of this collaboration has been the creation of a tool for the calculation of the interaction matrices. With this purpose I computed the respective covariance matrices out of the frequency
matrices provided by Gabriele Macari. I then obtained the first interaction matrices through the inversion of the covariance matrices and the application of an appropriate gauge invariance. After this first step the interaction matrices have been evaluated for their consistency. Despite the results obtained with this first inversion process are in good agreement with the chemical nature of the amino acids, the signal of the amino acid-fragment and fragment-fragment interactions is too weak. We therefore decided to analyse the effect on the effective matrices due to a change in the normalization of the frequency matrices. With this purpose Gabriele Macari built a novel database of connectivity networks that I used as starting point for the novel normalization procedure. In this database each pocket is represented by a graph in which the nodes are the elements composing the pocket, and the links are weighted by their distances. We thus reached the second step of the work (still in progress), that consists in the creation of the frequency matrices renormalized in blocks, starting from the interaction networks aforementioned. With this methodology the content of each pocket is evaluated with respect of the size of the pocket itself. The amino acid-amino acid, amino acid-fragment and fragment-fragment interactions are then normalized separately.
We did not yet test the interaction matrices because of the partial information of the firsts matrices I calculated. The work for the calculation of the new ones is close to completion and hence we will soon run
a benchmark on protein-ligand normally used as golden standard. In our tests we will assess the correlation of our scoring function against experimentally known binding affinities