The field of computational drug discovery and development has grown, with the aid of new computational tools for novel molecule discovery. In specific, generative deep learning models have excelled as tools to aid in navigating the large space of known molecules and in the creation of new molecules. These models are fed various representations of molecules as inputs and learn to perform a variety of things, such as the optimization of these molecules towards a targeted property. Ultimately, these generative learning models allow us to build bridges between chemical and continuous spaces to understand the compromise between invoking small incremental changes to radical modifications and generate optimal molecules for therapeutic benefits. The goal of this study pertains to creating a perturbation modeling framework in which we can conduct transformations of small molecules to SRC kinase inhibitors using a combination of a generative learning technique called Variational Auto-Encoders and Unsupervised Machine Learning techniques. This study focuses specifically using the methods above to transform molecules from various kinase inhibitor families to SRC Kinase Inhibitors. These generated molecules are evaluated using various physicochemical metrics and similarity metrics for validity of transformation. The results of this study demonstrate that Machine Learning based perturbation techniques can aid in the evolution of molecules from one chemical space to another. By combining Generative Learning frameworks with targeted based alteration of the continuous space, we allow for the emergence of novel molecules that are structurally adjacent to the various molecular scaffolds of SRC Kinase Inhibitors. Our findings also suggested that these vi molecules exhibit chemical and drug-likeness similarity to known SRC Kinase Inhibitors, displaying potential for synthetic and therapeutic benefits. Further studies should be conducted to determine whether the novel molecules that are generated contain potential to be synthesized for practical use.
","abstract_html":"<p>The field of computational drug discovery and development has grown, with the aid of new computational tools for novel molecule discovery. In specific, generative deep learning models have excelled as tools to aid in navigating the large space of known molecules and in the creation of new molecules. These models are fed various representations of molecules as inputs and learn to perform a variety of things, such as the optimization of these molecules towards a targeted property. Ultimately, these generative learning models allow us to build bridges between chemical and continuous spaces to understand the compromise between invoking small incremental changes to radical modifications and generate optimal molecules for therapeutic benefits. The goal of this study pertains to creating a perturbation modeling framework in which we can conduct transformations of small molecules to SRC kinase inhibitors using a combination of a generative learning technique called Variational Auto-Encoders and Unsupervised Machine Learning techniques. This study focuses specifically using the methods above to transform molecules from various kinase inhibitor families to SRC Kinase Inhibitors. These generated molecules are evaluated using various physicochemical metrics and similarity metrics for validity of transformation. The results of this study demonstrate that Machine Learning based perturbation techniques can aid in the evolution of molecules from one chemical space to another. By combining Generative Learning frameworks with targeted based alteration of the continuous space, we allow for the emergence of novel molecules that are structurally adjacent to the various molecular scaffolds of SRC Kinase Inhibitors. Our findings also suggested that these vi molecules exhibit chemical and drug-likeness similarity to known SRC Kinase Inhibitors, displaying potential for synthetic and therapeutic benefits. Further studies should be conducted to determine whether the novel molecules that are generated contain potential to be synthesized for practical use.</p>","abstract_has_math":false,"creators":["Krishnan, Keerthi"],"institution":null,"degree_name":"Master of Science (MS)","degree_level":"Thesis","degree_discipline":"Computational and Data Sciences","degree_department":null,"school":null,"contributors":["Gennady Verkhivker","Cyril Rakovski","Mohamed Allali"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2022,"date_issued":"2022-08-01T07:00:00Z","date_published":"2022-08-01T07:00:00Z","updated_at":"2026-07-24T01:38:24Z","subjects":["Machine Learning","Cheminformatics","Generative Learning","Protein Tyrosine Kinases","Kinase Inhibitors","Perturbation Modeling","Biochemistry","Biotechnology","Data Science","Molecular Biology","Pharmaceutics and Drug Design","Software Engineering"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://digitalcommons.chapman.edu/cads_theses/16","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Gennady Verkhivker","Cyril Rakovski","Mohamed Allali"]},{"key":"dc:creator","label":"Author","values":["Krishnan, Keerthi"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2024-06-24T07:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computational and Data Sciences"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science (MS)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Machine Learning","Cheminformatics","Generative Learning","Protein Tyrosine Kinases","Kinase Inhibitors","Perturbation Modeling","Biochemistry","Biotechnology","Data Science","Molecular Biology","Pharmaceutics and Drug Design","Software Engineering"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://digitalcommons.chapman.edu/cads_theses/16"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The field of computational drug discovery and development has grown, with the aid of new computational tools for novel molecule discovery. In specific, generative deep learning models have excelled as tools to aid in navigating the large space of known molecules and in the creation of new molecules. These models are fed various representations of molecules as inputs and learn to perform a variety of things, such as the optimization of these molecules towards a targeted property. Ultimately, these generative learning models allow us to build bridges between chemical and continuous spaces to understand the compromise between invoking small incremental changes to radical modifications and generate optimal molecules for therapeutic benefits. The goal of this study pertains to creating a perturbation modeling framework in which we can conduct transformations of small molecules to SRC kinase inhibitors using a combination of a generative learning technique called Variational Auto-Encoders and Unsupervised Machine Learning techniques. This study focuses specifically using the methods above to transform molecules from various kinase inhibitor families to SRC Kinase Inhibitors. These generated molecules are evaluated using various physicochemical metrics and similarity metrics for validity of transformation. The results of this study demonstrate that Machine Learning based perturbation techniques can aid in the evolution of molecules from one chemical space to another. By combining Generative Learning frameworks with targeted based alteration of the continuous space, we allow for the emergence of novel molecules that are structurally adjacent to the various molecular scaffolds of SRC Kinase Inhibitors. Our findings also suggested that these vi molecules exhibit chemical and drug-likeness similarity to known SRC Kinase Inhibitors, displaying potential for synthetic and therapeutic benefits. Further studies should be conducted to determine whether the novel molecules that are generated contain potential to be synthesized for practical use.
"]},{"key":"dc:source","label":"Dc Source","values":["K. Krishnan, \"Perturbation modeling for molecular design of protein tyrosine kinase inhibitors using unsupervised machine learning,\" M. S. thesis, Chapman University, Orange, CA, 2022. https://doi.org/10.36837/chapman.000397"]},{"key":"dc:title","label":"Title","values":["Perturbation Modeling for Molecular Design of Protein Tyrosine Kinase Inhibitors using Unsupervised Machine Learning"]}]}],"canonical_facts":{"dc:contributor":["Gennady Verkhivker","Cyril Rakovski","Mohamed Allali"],"dc:creator":["Krishnan, Keerthi"],"dc:date.available":["2024-06-24T07:00:00Z"],"dc:description.abstract":["The field of computational drug discovery and development has grown, with the aid of new computational tools for novel molecule discovery. In specific, generative deep learning models have excelled as tools to aid in navigating the large space of known molecules and in the creation of new molecules. These models are fed various representations of molecules as inputs and learn to perform a variety of things, such as the optimization of these molecules towards a targeted property. Ultimately, these generative learning models allow us to build bridges between chemical and continuous spaces to understand the compromise between invoking small incremental changes to radical modifications and generate optimal molecules for therapeutic benefits. The goal of this study pertains to creating a perturbation modeling framework in which we can conduct transformations of small molecules to SRC kinase inhibitors using a combination of a generative learning technique called Variational Auto-Encoders and Unsupervised Machine Learning techniques. This study focuses specifically using the methods above to transform molecules from various kinase inhibitor families to SRC Kinase Inhibitors. These generated molecules are evaluated using various physicochemical metrics and similarity metrics for validity of transformation. The results of this study demonstrate that Machine Learning based perturbation techniques can aid in the evolution of molecules from one chemical space to another. By combining Generative Learning frameworks with targeted based alteration of the continuous space, we allow for the emergence of novel molecules that are structurally adjacent to the various molecular scaffolds of SRC Kinase Inhibitors. Our findings also suggested that these vi molecules exhibit chemical and drug-likeness similarity to known SRC Kinase Inhibitors, displaying potential for synthetic and therapeutic benefits. Further studies should be conducted to determine whether the novel molecules that are generated contain potential to be synthesized for practical use.
"],"dc:identifier":["https://digitalcommons.chapman.edu/cads_theses/16"],"dc:source":["K. Krishnan, \"Perturbation modeling for molecular design of protein tyrosine kinase inhibitors using unsupervised machine learning,\" M. S. thesis, Chapman University, Orange, CA, 2022. https://doi.org/10.36837/chapman.000397"],"dc:subject":["Machine Learning","Cheminformatics","Generative Learning","Protein Tyrosine Kinases","Kinase Inhibitors","Perturbation Modeling","Biochemistry","Biotechnology","Data Science","Molecular Biology","Pharmaceutics and Drug Design","Software Engineering"],"dc:title":["Perturbation Modeling for Molecular Design of Protein Tyrosine Kinase Inhibitors using Unsupervised Machine Learning"],"thesis:degree_discipline":["Computational and Data Sciences"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["Master of Science (MS)"]},"updated_at":"2026-07-24T01:38:24Z"}