I'm a computational biologist/engineer/bioinformatician with several years experience with high-dimensional data analysis. I have a penchant for programming and programming language design with some hardware and entrepreneurial experience mixed in. I also have a wide range of additional interests and hobbies (cryptocurriencies, biohacking, etc.).
I am currently finishing my Ph.D. at Texas A&M University in the Genomic Signal Processing laboratory where I apply tools from modern engineering, mathematics, and statistics in order to understand and treat cancer.
Specifically, I have worked closely with biologists to develop Bayesian statistical models to uncover multivariate interactions among genes from RNA-Seq data. This allows us to classify types of cancer better than nonlinear SVM methods (which themselves are typically considered state of the art). I have also modeled biological regulatory networks using probabilistic Markov models that incorporate pathway knowledge from the biology literature as well as high-throughput data.
In the future, I want to continue using my knowledge and skills to enable principled approaches towards the detection, classification, and treatment of human diseases.
Every single one of the 40 trillion cells in your body has tens of thousands of genes which are coordinated in a chaotic symphony of regulation for your continued living (homeostasis in biological parlance). For decades now, biologists have been studying this regulatory network piece by piece through a variety of ingenious techniques and a mind-boggling amount of effort.
The trouble is, this biological pathway knowledge is incomplete and sometimes contradictory. Due to this, building predictive models from this knowledge is difficult. In this work, we used a modeling approach embracing the uncertainty through a family of Markov chains (which themselves are a random dynamical system model).
Specifically, we obtained pathways from the biological literature regarding the NF-kB network which is central to cellular inflammatory processes. Then constructing a family of Markov chains from these pathways, we could then make predictions of how the network might evolve under the deletion of several genes under the effect of various stimuli. These predictions were then qualitatively validated against additional biological literature where these experiments were carried out on mouse gene knockout models.
Texas A&M University
Advisor: Edward Dougherty
May 2015 (Expected)
Texas A&M University
Minor: Mathematics (summa cum laude)