Algorithmic Identification of Murzymes and Murburn Mechanisms Based on Structural, Theoretical, Experimental, and Generic Features

TL;DR

This paper introduces algorithms to identify murzymes and murburn mechanisms in heme-protein systems using structural and experimental data.

Contribution

A novel algorithmic framework using AI-ML and Ockham's razor to classify murzymes with high accuracy.

Findings

An eight-point comparison showed murburn model outperforms classical models in heme-protein systems.

AI-ML methods achieved total accuracy in delineating murzymes using 20 parameters.

A web-based portal and LLM-SVM model were developed for murzyme classification with ~84% accuracy.

Abstract

Murburn concept is a decade‐old theorization for explaining several cellular redox‐metabolic and electro/mechanophysiological outcomes. It stems from “mured burning”, connoting delocalized stochastic electron‐transfer processes, involving diffusible reactive species (DRS). The thermodynamic–kinetic–mechanistic (TKM) aspects of several cellular activities are seamlessly woven with the idea of murburn, also recently consolidated with quantitative models. Therefore, it is now opportune to algorithmically delineate the various features of protein/metabolic systems that generate, modulate/stabilize, or utilize DRS. Herein, we meta‐analyzed three simple/single heme‐protein systems: (1) extracellular heme‐haloperoxidase, (2) membrane‐bound cyclooxygenase, and (3) soluble hemoglobin; and also two complex/multiprotein systems (incorporating heme‐enzymes): (4) the hepatocyte xenobiotic metabolism proteins on endoplasmic reticulum, and (5) the mitochondrial oxidative phosphorylation machinery. We first tabulated an eight‐point comparison of the classical mechanistic models in literature with the murburn model for these five systems and also performed an internal consistency check for the classical models. Thereafter, we compared eight distinct postulates of both models for these systems and employed an Ockham′s razor‐based algorithm for the preferred mechanistic model. Further, based on 20 structural, theoretical, and experimental parameters, we employed decision‐tree/random‐forest (AI‐ML) methods to delineate murzymes (murburn systems) with total accuracy. Furthermore, we developed a simple web‐based portal for mechanistic parsing of murzymes from classical enzymes. Using the classification thus derived, we developed an LLM‐SVM–based model to demarcate murzymes (with ~84% accuracy), parameterizing only text‐descriptors in PDB (RCSB) files. Herein, we also provide a brief projection of how these novel mechanistic and algorithmic analyses impact research in redox/TKM‐enzymology.

Figures