Tutorial SB06—Machine Learning for Peptide-Based Materials Design

Instructors: Tell Tuttle, University of Strathclyde; Alexander van Teijlingen, University of Strathclyde; Dhwanit Dave, Advanced Science Research Center, City University of New York

The first half of the Day will involve three lectures that focus on the theory that underlies the methods that will be explored in the second half of the days tutorials. These include a tutorial lecture on Coarse Grain Molecular Dynamics and how it can be used to predict Materials properties; a tutorial lecture on Machine learning in Materials design with a focus on the type of neural networks commonly used and the metrics that can be used to assess the quality of the results; a tutorial lecture on the analysis of the outcomes of simulations and how different metrics from simulations can be related to macroscale properties of the materials.

The second half of the day will focus on a practical approach where the participants will be asked to work through examples of the three topics on their laptops. We will contact registered participants in advance of the session to let them know what software should be downloaded and where they can access files etc.

Learning Objectives:

(i) Identify whether a method is coarse grain or atomistic;

(ii) State the most common parameters that need to be defined in a coarse grain simulation; (iii) Transform an atomistic representation of a peptide into a coarse grain representation;

(iv) Run a coarse grain molecular dynamics simulation;

(v) Differentiate between active and passive machine learning;

(vi) Quantify the quality of a machine learning method using different metrics;

(vii) Run a machine learning algorithm across various sized datasets;

(viii) Evaluate the outcomes of the machine learning program runs to determine the quality of the algorithm;

(ix) Define various metrics such as SASA, AP score, H-bond measures, Rg, etc.

(x) Visualise the trajectory of a CG run;

(xi) Manipulate the visualization modes of CG trajectory to highlight different features;

(xii) Calculate various metrics based on the trajectories.

Part I Programs Required

Part II Programs Required

Tutorial Schedule