I am motivated by the questions: “why do some drugs work and why can’t we engineer better performance at initial drug design?” and I endeavor to discover models that can answer these questions. My career goal is to be an independent scientist engineering models for the academic-translational interface. My scientist-engineer approach affords me a perspective that is both inquisitive – “what are the unknowns about the systems?” – and creative – “how can we apply this knowledge to advance human health?”, that will be uniquely beneficial to understanding drug and disease systems.
Here you'll find an expanded view of my curriculum vitae with some more motivation behind the work.
Computational methods discern drug effects without prior mechanistic understanding
For my senior, undergraduate thesis, I created a combined experimental and computational pipeline for identifying mechanisms for various pro-angiogenic compounds. Angiogenesis, or blood vessel growth, is an important feature of wound healing and promoting this process is desirable in tissue engineering applications. Through a collaborator, the lab was testing multiple chemical compounds with known effects on blood vessel development. However, we did not have mechanistic understanding of how these compounds affected blood vessel development nor know how these compounds differed in their effects. I leveraged my training in systems biology to design an experimental pipeline to measure mRNA expression in cells exposed to these compounds and a modeling pipeline for discerning key signals in response to treatment. I designed this project and collected data and a senior postdoc prepared the manuscript. This project was innovative because very few researchers had applied systems biology modeling to tissue engineering applications and the project moved the field vertically by demonstrating a path forward for identifying proangiogenic drug mechanisms. This project demonstrated the power of computation for therapeutics, but I learned that more complex methods would be necessary for understanding drug mechanisms.
Computational network methods create cohesive models of gene-interference screens and uncover new biology
In my PhD work, I pursed a deeper understanding of computational, network models, with an emphasis on their utility for gene-interference screens. Gene-interference technologies, such as rna-interference (RNAi) became popular because they could simultaneously perturb the function of several genes to understand which genes affected a particular biological phenotype. However, technical challenges, such as seed-mismatching, caused single RNAi to affect unintended genes and cause “off-target effects”. These technical challenges made it difficult to discern which genes were truly affecting a biological phenotype. I discovered that network methods could provide context for which genes were affecting a phenotype of interest. These methods leveraged known gene-gene interactions to create a pathway representation of experimental data, essentially using a “guilt-by-association” perspective to identify which genes were affecting a phenotype. I pursued this approach in two cancer settings – acute lymphoblastic leukemia (ALL) and transforming growth factor alpha (TGFalpha) ectodomain shedding for growth-driven cancers. After model construction, I validated model predictions using both in vitro and in vivo experiments. In ALL, I discovered that WW domain containing E3 ubiquitin protein ligase 1 (WWP1) had an in vivo-specific role in ALL progression. In TGFalpha ectodomain shedding, I discovered that interleukin 1 receptor associated kinase 1 (IRAK1) had a previously uncharacterized role on growth factor shedding and that existing IRAK1 inhibitors could perturb this system if used therapeutically. These projects advanced the field because they demonstrated that computational methods could discover novel insights from preexisting RNAi screening data. The work was innovative because it used uncommon analysis techniques to develop new hypotheses from RNAi screens. I gained an appreciation for complex, systems-level models and expertise in developing and adapting these models to multiple biological systems.
Drug pathway models contain associations to safety risks and efficacy phenotypes.
As a fellow at the UCSF-Stanford Center of Excellence in Regulatory Science and Innovation (CERSI), I explored the question “What relevant safety signals exist for drug targets in development?”. I hypothesized that interaction pathways around a drug’s targets contained information relevant for understanding safety signals, such as associations to adverse event phenotypes. Using my experience in pathways modeling, I created PathFX, an algorithm that first identifies a putative pharmacodynamic drug pathway and then analyzes biological phenotypes significantly associated with that pathway. I analyzed over 200 marketed drugs and discovered that the same pathway formalism applied to a specific drug could identify both efficacy phenotypes (i.e. the drug’s marketed disease indication) and safety phenotypes (i.e. safety risks listed on the drug label). During my ORISE fellowship, I brought the platform to the US Food and Drug Administration (FDA) and tested the ability to anticipate safety events for drug targets in development. Through this collaboration, I have refined the model and created a webserver to enable continued use of the PathFX formalism at the FDA and within the research community. The project is innovative because it leveraged a set of protein-protein interactions to explain how a drug’s target(s) pathway is associated with both efficacy and safety phenotypes. The work advanced the field because it provided a formalism for considering drug target associations beyond the targets themselves, and because the platform will be broadly accessible to computation and non-computational researchers. In addition to safety and efficacy phenotypes, PathFX also identified additional disease phenotypes for which a given drug may or may not be used. These associations represented potential repurposing opportunities, and inspired me to pursue more detailed pathway models with an aim for identifying drug targets with strong pathways relationships to diseases of interest.
Here you'll find an expanded view of my curriculum vitae with some more motivation behind the work.
Computational methods discern drug effects without prior mechanistic understanding
For my senior, undergraduate thesis, I created a combined experimental and computational pipeline for identifying mechanisms for various pro-angiogenic compounds. Angiogenesis, or blood vessel growth, is an important feature of wound healing and promoting this process is desirable in tissue engineering applications. Through a collaborator, the lab was testing multiple chemical compounds with known effects on blood vessel development. However, we did not have mechanistic understanding of how these compounds affected blood vessel development nor know how these compounds differed in their effects. I leveraged my training in systems biology to design an experimental pipeline to measure mRNA expression in cells exposed to these compounds and a modeling pipeline for discerning key signals in response to treatment. I designed this project and collected data and a senior postdoc prepared the manuscript. This project was innovative because very few researchers had applied systems biology modeling to tissue engineering applications and the project moved the field vertically by demonstrating a path forward for identifying proangiogenic drug mechanisms. This project demonstrated the power of computation for therapeutics, but I learned that more complex methods would be necessary for understanding drug mechanisms.
- Sefcik LS, Wilson JL, Papin JA, Botchwey EA. Harnessing systems biology approaches to engineer functional microvascular networks. Tissue Eng Part B Rev. 2010 Jun;16(3):361-70. PubMed PMID: 20121415; PubMed Central PMCID: PMC2946904. PDF Link
- Das A, Merrill P, Wilson JL, Freshcorn B, Paige M, Capitosti S, Brown M, Turner T, Sok MP, Song H, Botchwey EA. Evaluating angiogenic potential of small molecules using genetic network approaches. Regenerative Engineering and Translational Medicine. 2019: 1-12. PDF Link
Computational network methods create cohesive models of gene-interference screens and uncover new biology
In my PhD work, I pursed a deeper understanding of computational, network models, with an emphasis on their utility for gene-interference screens. Gene-interference technologies, such as rna-interference (RNAi) became popular because they could simultaneously perturb the function of several genes to understand which genes affected a particular biological phenotype. However, technical challenges, such as seed-mismatching, caused single RNAi to affect unintended genes and cause “off-target effects”. These technical challenges made it difficult to discern which genes were truly affecting a biological phenotype. I discovered that network methods could provide context for which genes were affecting a phenotype of interest. These methods leveraged known gene-gene interactions to create a pathway representation of experimental data, essentially using a “guilt-by-association” perspective to identify which genes were affecting a phenotype. I pursued this approach in two cancer settings – acute lymphoblastic leukemia (ALL) and transforming growth factor alpha (TGFalpha) ectodomain shedding for growth-driven cancers. After model construction, I validated model predictions using both in vitro and in vivo experiments. In ALL, I discovered that WW domain containing E3 ubiquitin protein ligase 1 (WWP1) had an in vivo-specific role in ALL progression. In TGFalpha ectodomain shedding, I discovered that interleukin 1 receptor associated kinase 1 (IRAK1) had a previously uncharacterized role on growth factor shedding and that existing IRAK1 inhibitors could perturb this system if used therapeutically. These projects advanced the field because they demonstrated that computational methods could discover novel insights from preexisting RNAi screening data. The work was innovative because it used uncommon analysis techniques to develop new hypotheses from RNAi screens. I gained an appreciation for complex, systems-level models and expertise in developing and adapting these models to multiple biological systems.
- Wilson JL, Hemann MT, Fraenkel E, Lauffenburger DA. Integrated network analyses for functional genomic studies in cancer. Semin Cancer Biol. 2013 Aug;23(4):213-8. PubMed PMID: 23811269; PubMed Central PMCID: PMC3844556. PDF Link
- Wilson JL, Dalin S, Gosline S, Hemann M, Fraenkel E, Lauffenburger DA. Pathway-based network modeling finds hidden genes in shRNA screen for regulators of acute lymphoblastic leukemia. Integr Biol (Camb). 2016 Jul 11;8(7):761-74. PubMed PMID: 27315426; PubMed Central PMCID: PMC5224708. PDF Link
- Wilson JL, Kefaloyianni E, Stopfer L, Harrison C, Sabbisetti VS, Fraenkel E, Lauffenburger DA, Herrlich A. Functional Genomics Approach Identifies Novel Signaling Regulators of TGFα Ectodomain Shedding. Mol Cancer Res. 2018 Jan;16(1):147-161. PubMed PMID: 29018056; PubMed Central PMCID: PMC5859574. PDF Link
Drug pathway models contain associations to safety risks and efficacy phenotypes.
As a fellow at the UCSF-Stanford Center of Excellence in Regulatory Science and Innovation (CERSI), I explored the question “What relevant safety signals exist for drug targets in development?”. I hypothesized that interaction pathways around a drug’s targets contained information relevant for understanding safety signals, such as associations to adverse event phenotypes. Using my experience in pathways modeling, I created PathFX, an algorithm that first identifies a putative pharmacodynamic drug pathway and then analyzes biological phenotypes significantly associated with that pathway. I analyzed over 200 marketed drugs and discovered that the same pathway formalism applied to a specific drug could identify both efficacy phenotypes (i.e. the drug’s marketed disease indication) and safety phenotypes (i.e. safety risks listed on the drug label). During my ORISE fellowship, I brought the platform to the US Food and Drug Administration (FDA) and tested the ability to anticipate safety events for drug targets in development. Through this collaboration, I have refined the model and created a webserver to enable continued use of the PathFX formalism at the FDA and within the research community. The project is innovative because it leveraged a set of protein-protein interactions to explain how a drug’s target(s) pathway is associated with both efficacy and safety phenotypes. The work advanced the field because it provided a formalism for considering drug target associations beyond the targets themselves, and because the platform will be broadly accessible to computation and non-computational researchers. In addition to safety and efficacy phenotypes, PathFX also identified additional disease phenotypes for which a given drug may or may not be used. These associations represented potential repurposing opportunities, and inspired me to pursue more detailed pathway models with an aim for identifying drug targets with strong pathways relationships to diseases of interest.
- Wilson JL, Altman RB. Biomarkers: Delivering on the expectation of molecularly driven, quantitative health. Exp Biol Med (Maywood). 2018 Feb;243(3):313-322. PubMed PMID: 29199461; PubMed Central PMCID: PMC5813871. PDF Link
- Wilson JL. A scientist engineer’s contribution to therapeutic discovery. Experimental Biology and Medicine. Experimental Biology and Medicine. 2018; 0: 1–8. PDF Link
- Wilson JL, Racz R, Liu T, Adeniyi O, Sun J, Ramamoorthy A, Pacanowski M, Altman RB. PathFX provides mechanistic insights into drug efficacy and safety for regulatory review and therapeutic development. PLoS computational biology. PLoS Comput Biol. 2018; 14(12): e1006614. PDF Link
- Wilson, J.L., M. Wong, A. Chalke, N. Stepanov, D. Petkovic, R.B. Altman. “PathFXweb: a web application for identifying drug safety and efficacy phenotypes.” Bioinformatics. 2019 May 22. pii: btz419. doi: 10.1093/bioinformatics/btz419. [Epub ahead of print]. PDF link Supplement Link