The Harmonizome is a comprehensive resource that integrates and curates data on genes and proteins from over 70 major online resources. It aims to address the need for better knowledge extraction and data reuse by organizing and mining information from various omics studies and biomedical literature. The Harmonizome collects and processes data into approximately 72 million functional associations between genes/proteins and their attributes, such as physical interactions, expression in cell lines and tissues, genetic associations, and changes in expression after drug treatment. These associations are stored in a relational database with rich metadata. The Harmonizome provides a web portal with a graphical user interface, a web service, and a mobile app for querying, browsing, and downloading the collected data. The resource enables researchers to discover novel relationships between biological entities and form data-driven hypotheses for experimental validation. The Harmonizome also includes machine learning applications to predict novel properties of genes and proteins, such as ion channels, mouse phenotypes for single gene knockouts, endogenous ligands for G-protein-coupled receptors, and kinase substrates.The Harmonizome is a comprehensive resource that integrates and curates data on genes and proteins from over 70 major online resources. It aims to address the need for better knowledge extraction and data reuse by organizing and mining information from various omics studies and biomedical literature. The Harmonizome collects and processes data into approximately 72 million functional associations between genes/proteins and their attributes, such as physical interactions, expression in cell lines and tissues, genetic associations, and changes in expression after drug treatment. These associations are stored in a relational database with rich metadata. The Harmonizome provides a web portal with a graphical user interface, a web service, and a mobile app for querying, browsing, and downloading the collected data. The resource enables researchers to discover novel relationships between biological entities and form data-driven hypotheses for experimental validation. The Harmonizome also includes machine learning applications to predict novel properties of genes and proteins, such as ion channels, mouse phenotypes for single gene knockouts, endogenous ligands for G-protein-coupled receptors, and kinase substrates.