Evaluación crítica de modelos de lipofilicidad y solubilidad en moléculas pequeñas: un enfoque con aprendizaje automático
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
La lipofilicidad y la solubilidad son propiedades fisicoquímicas interrelacionadas y con alta relevancia en el diseño de fármacos para determinar la biodisponibilidad de moléculas. La lipofilicidad, (log 𝑃N, o log 𝐷pH en sistemas dependientes del pH) se ha asociado con propiedades farmacocinéticas como la permeabilidad en membranas, toxicidad, bioacumulación, etc. La solubilidad intrínseca (log 𝑆0) se asocia con la absorción y distribución de fármacos en el cuerpo. Asimismo, se han encontrado aplicaciones recientes de estas propiedades en áreas como la química de alimentos, ambiental y en biología computacional. Por ello, en los últimos años, ha sido de gran interés la creación de modelos computacionales (basados en razonamientos deductivos e inductivos) para la predicción adecuada de log 𝐷pH y log 𝑆.Este proyecto analiza el poder predictivo de dos ecuaciones derivadas termodinámicamente para la determinación del log 𝐷pH en sistemas n-octanol/agua, donde se discute la importancia de la consideración de especies iónicas en la fase orgánica. Similarmente, se evalúa la Ecuación General de la Solubilidad (GSE) utilizada para el cálculo de log 𝑆 de moléculas pequeñas basada en el punto de fusión y el coeficiente de partición. Asimismo, se plantea la creación de modelos de aprendizaje automático para indicar, con base en características estructurales de las moléculas, en cuáles casos las ecuaciones de log 𝐷pH y log 𝑆 pueden aplicarse para obtener predicciones confiables. Con ello, se pretende crear una guía para la comunidad científica que trabaje con moléculas pequeñas en el diseño de fármacos, química de alimentos y ambiental, sobre cuál perfil de lipofilicidad y solubilidad es el más apropiado en sus respectivas investigaciones. Finalmente, se plantea la implementación de estos modelos en programas para sus moléculas de interés.
Lipophilicity and solubility are interrelated physicochemical properties of high relevance in drug design for determining molecular bioavailability. Lipophilicity (determined by the partition coefficient, log 𝑃N, or log 𝐷pH in pH-dependent systems) is associated with pharmacokinetic properties such as membrane permeability, toxicity, and bioaccumulation. Solubility (log 𝑆0) governs drug absorption and distribution in the body.Recent applications of these properties have also emerged in fields such as food chemistry, environmental chemistry, and computational biology. Consequently, in recent years, significant interest has developed in creating computational models (based on deductive and inductive reasoning) for the accurate prediction of log 𝐷pH and log 𝑆. This project analyzes the predictive power of two thermodynamically derived equations for determining log 𝐷pH in n-octanol/water systems, discussing the critical importance of considering ionic species in the organic phase. Similarly, the General Solubility Equation (GSE), used to calculate the log 𝑆 of small molecules based on melting point and partition coefficient, is evaluated. Furthermore, the development of machine learning models is proposed to indicate, based on molecular structural features, in which cases the log 𝐷pH and log 𝑆 equations can be applied to yield reliable predictions. The aim is to create a guideline for the scientific community working with small molecules in drug design, food, and environmental chemistry, advising on the most appropriate lipophilicity and solubility profiles for their respective research. Finally, the implementation of these models into accessible software is proposed, enabling any interested user to determine the best log 𝐷pH and log 𝑆 models for their molecules of interest.
Lipophilicity and solubility are interrelated physicochemical properties of high relevance in drug design for determining molecular bioavailability. Lipophilicity (determined by the partition coefficient, log 𝑃N, or log 𝐷pH in pH-dependent systems) is associated with pharmacokinetic properties such as membrane permeability, toxicity, and bioaccumulation. Solubility (log 𝑆0) governs drug absorption and distribution in the body.Recent applications of these properties have also emerged in fields such as food chemistry, environmental chemistry, and computational biology. Consequently, in recent years, significant interest has developed in creating computational models (based on deductive and inductive reasoning) for the accurate prediction of log 𝐷pH and log 𝑆. This project analyzes the predictive power of two thermodynamically derived equations for determining log 𝐷pH in n-octanol/water systems, discussing the critical importance of considering ionic species in the organic phase. Similarly, the General Solubility Equation (GSE), used to calculate the log 𝑆 of small molecules based on melting point and partition coefficient, is evaluated. Furthermore, the development of machine learning models is proposed to indicate, based on molecular structural features, in which cases the log 𝐷pH and log 𝑆 equations can be applied to yield reliable predictions. The aim is to create a guideline for the scientific community working with small molecules in drug design, food, and environmental chemistry, advising on the most appropriate lipophilicity and solubility profiles for their respective research. Finally, the implementation of these models into accessible software is proposed, enabling any interested user to determine the best log 𝐷pH and log 𝑆 models for their molecules of interest.
Description
Keywords
Física Química, Aprendizaje Automático, Farmacología, Química de alimentos, Química ambiental, Propiedad química