UC San Diego

Genomic databases help in the understanding of how genetic changes alter a phenotype and herein are linked to a disease. Similar resources are needed to study small molecules; they can alter a target gene’s normal activity (e.g., drugs, ligands, inhibitors). But, contrary to genomic databases, chemogenomic databases are limited and less annotated. Furthermore, molecule identifiers are not standardized and there are variations in naming depending on the database. This poses a challenge when trying to identify how a drug interacts with its biological target and whether it results in the desired phenotypic response. Knowledge on drugs’ mechanism of action (MoA) or biological target helps translating those candidates into therapeutic treatments. In this dissertation, we use in silico methods to support the translation of genomic data into effective new drugs, by analyzing chemical libraries for early identification of research efforts that uses previously used compounds, close analogs or existing literature; and drug targets/MoA predictions. We find that by chemical querying and chemical comparisons we could provide target hypothesis for scaffold families and, paired with validation studies (e.g., in vitro evolution), it results in faster and more efficient drug development.In Chapter 1, we describe CACTI, a tool that explores the chemical and biological space pragmatically, to provide an understanding of chemical structures and their impact on a biological system. This sets the basis for querying, comparing, clustering, and predicting potential targets for chemical compounds. In Chapter 2, we demonstrate that small molecules could have liabilities. Through the analysis of 724 Plasmodium falciparum genomes resistant to one of 118 small molecules, we show commonalities between resistant clones setting the foundation for machine learning predictive approaches. Chapter 3 describes the use of data mining and protein structure prediction to identify novel antimalarial drug targets, supporting the development of drugs that differ in MoA and resistance liabilities. Lastly, in Chapter 4, we explore the use of single-cell sequencing to provide clues about the mechanism for uncharacterized chemotypes. Contrasting transcriptomic profiles of P. falciparum parasites and drug-treated parasites, we found upregulation of biological pathways for atovaquone and artemisinin.