Artificial Intelligence Model Based on Algebraic Topology for Protein Structure Analysis and Prediction

Abstract

With the availibility and the easy access to protein data through different publicaly available databases a lot of questions are raised on how to make sense from data in aim to figure out new strategies in reproducing meaningfull conclusions that can anticipate in building a consistent theoretical knowledge in the field of protein structure prediction and analysis; and regarding the nature of a metric in biology and emphasizing on its behaviour as a similarity measure we are presenting a model built on the assumption that only the shape of data can tell about the data; the learning approach is derived from algebraic topology, We will precisely be showing how our quotioned spaces could qualititavely give insight into how building good homomorphisms can help identifying accurate neural networks, by encoding the two first homologies H₁ to H₀ using a boundary operator, the algorithms are originated from algebraic geometry Basically two main algorithms are used the Buchberger's algorithm and Shreyer's algorithm.