A novel random forest approach to predict phase transition

Machine learning algorithms design the model that predicts the system’s behaviour concerning the data used as knowledge. Some application areas and biopolymers have used machine learning to solve complex computational problems. The phase transition from one structure to another with change in an environment like increasing temperature or pH value is observed in many biopolymers. This phenomenon related to change in the biopolymer structure with change in environment has been widely studied, as it plays a significant role in identifying the properties of biopolymers related to conformations of a molecular structure. This paper explores various machine learning algorithms to develop a model that can predict the biopolymer structure. Apart from structure prediction, a novel algorithm is proposed that extends the model in determining the phase transition in the given biopolymer. This algorithm merges two machine learning concepts where a classification algorithm (like random forest) with high accuracy is used to predict the structure of the given biopolymer at varying temperatures, and a segmented model of breakpoint analysis determines that phase transition occurs with a temperature change. The breakpoint is an abrupt change representing the transition between states during the variation in time series data. R studio is used to create the model, and two different biopolymers, namely L-alanine and L-cysteine, with changes in physical condition like temperature are used for validation. The result of the experiment proves that the proposed algorithm significantly improves the detection of phase transition in biopolymers.