Nitya Thakkar

Brown University, Sc.B. Computer Science, June 2023
Activities: Meiklejohn Peer Advisor, Women in CS mentor, Brown Elementary Afterschool Mentoring volunteer, The Brown Daily Herald staff writer, Brown Abhinaya - classical Indian dance team (co-captain 2021-22), Full Stack at Brown software engineer, CS Departmental Undergraduate Group Leader, Brown Machine Intelligence Community content creator and instructor
TA Experience: Head TA for Deep Learning Spring '23, where I oversee a TA staff of 25. TA for Deep Learning, Fall 2022 and Spring 2022; Introduction to Computer Systems, Fall 2021; Linear Algebra, Spring 2021. Responsibilities include course development, grading problem sets/projects, and holding weekly office hours.
St. Paul Academy and Summit School, June 2019

Coursework at Brown

Computer Science: Deep Learning, Machine Learning, Computational Molecular Biology, Data Science, Computer Vision, Introduction to OOP and Computer Science, Introduction to Algorithms and Data Structures, Computer Systems
Math: Multivariable Calculus, Linear Algebra, Discrete Structures and Probability, Statistical Inference and Probability
Other: Genetics, Biochemistry, Chemistry, Organic Chemistry

Programming Skills

Proficient: Python, Java, C, PyTorch, and TensorFlow
Experience with: HTML/CSS, JavaScript, React, R, x86-64 Assembly Language, LaTeX, and Git

Awards and Accomplishments

Computing Research Association's (CRA) Outstanding Undergraduate Researcher Award 2022 Honorable Mention : This prestigious award recognizes undergraduate students who show outstanding research potential in an area of computing research.

Microsoft Research, Biomedical ML
Dr. Kevin Yang, May 2022 - Present
I created a denoising diffusion probabilistic model to generate 2D protein alignments (Multiple Sequence Alignments). I hypothesized that Order Agnostic Autoregressive Diffusion Model in a two-dimensional space on Multiple Sequence Alignments (MSAs) will be able to capture the natural variation in the protein sequences and generate new proteins with desired structures and functions. I then evaluated the quality of the designed protein alignments by assessing pairwise sequence similarity, rates of pairwise amino acid substitutions, and measuring how well the secondary structures are encoded for protein engineering tasks; this can enable targeted protein therapies.

Brown University, Computational Biology Lab
Dr. Ritambhara Singh, Jan. 2020 - Present

Honors senior thesis project aims to characterize the glioblastoma cellular environment using gene expression and cell state energy data; proposing a novel methodology for Bayesian inference on graphs using deep learning approaches.

Worked on a project to predict gene expression values in glioblastomas, which is cancer in the brain that is characterized by its aggressive nature. My aim was to predict these gene expression values from other epigenetic data to better understand the genetic mechanisms in brain tumors. We found that a variety of baseline machine learning models, such as support vector machines and random forest models did not learn many significant relationships. I trained a Convolutional Neural Network (CNN) to accomplish this task.

Predicting A/B compartments from histone modifications using deep learning
Co-first author on a project to predict three-dimensional (3D) organization of our DNA from one-dimensional (1D) biological experiments to understand gene interaction mechanisms. Understanding 3D spatial DNA organization is critical to understanding gene interactions; however, it is currently difficult and expensive to manually obtain 3D data. I hence helped create a recurrent neural network (RNN) to predict the 3D DNA organization from easily available 1D information about biological factors. This model outperformed the mean baseline model across all cell lines, which indicates it is able to accurately predict the 3D interactions. I also implemented and ran baseline methods to compare the accuracy of my model, and improved the data pre-processing pipeline to increase the resolution of the data, which allowed us to view regions of the genome more precisely. Languages used: Python, R, Git

Research Assistant at the Broad Institute of MIT and Harvard
Dr. Neriman Tokcan, June 2021 - Dec. 2021

I was selected for the prestigious Broad Summer Research Program (2021), where I conducted research in Dr. Todd Golub’s and Dr. Caroline Uhler’s labs under the mentorship of Neriman Tokcan, Ph.D. I worked with the lymphatic system cancer, Classical Hodgkin’s Lymphoma, which is unique since it has a diverse composition of tumor cells and immune cells. Previous studies have found these tumor cells are dependent on their environment for survival; however, not much is known about what specific cells and genes contribute to their survival. Therefore, the goal of my project was to develop new computational methods to analyze spatial data that captures the tumor cell environment. I implemented a feed-forward neural network to identify the types of cells in spatial data from gene expression data. I also created a new method to analyze the spatial relationship and interactions between cells in the microenvironment. I anticipate these methods will facilitate the downstream development of more targeted cancer therapies if wet-lab researchers target specific immune cells that support the survival of cancerous cells. I presented this project at the Annual Biomedical Research Conference for Minority Students in November. Languages used: Python, R

University of Massachusetts - Amherst, Food Science Lab
Dr. Yeonhwa Park, July 2018 - Aug. 2018

The goal of my project was to determine the effects of Sulforaphane, a compound found in cruciferous vegetables, on aging, obesity and oxidative stress in Caenorhabditis elegans. Previous studies have found it to have anticarcinogenic effects, and I sought to determine if it could have other positive effects on health. I independently ran PCR analyses on specific genes, and conducted various wet-lab assays. The results indicated that Sulforaphane may improve oxidative stress response in C. elegans through the post-translational regulation of SKN-1, which indicates that it may slow aging. I received awards for both my poster presentation and paper at the Twin Cities Regional Science Fair and Minnesota State Science Fair. I also qualified as a finalist for the prestigious Intel International Science and Engineering Fair, which I attended in May 2019. Through this project, I made extensive use of different data analysis processes using Excel, Statistical Analysis Software (one-way ANOVA and chi-square test), GraphPad Prism (Logrank test) and SPSS.