About Me
Hi, I’m Rana Barghout, a PhD student at the University of Toronto and Vector Institute, working under the supervision of Krishna Mahadevan (Laboratory for Metabolic Systems Engineering) and Benjamin Sanchez-Lengeling (Chemical Cognition Lab).
My research focuses on building systems that predict the functional parameters of compound-protein interactions, deploying them within genome-scale, constraint-based metabolic models, and probing whether these models are learning real biological signal, from individual residues all the way up to whole cellular systems. I’m also interested in graph learning, and have applied graph neural networks to problems ranging from molecules and protein-protein interactions to genome-scale metabolic networks.
Research Interests
- Molecular Modeling
- Cell systems
- Personalized Medicine
- Drug Discovery
- Deep Learning
- Graph Learning
- Neural Network-based Linear Programming
Outside of Research
When I’m not in front of a computer, I enjoy road cycling, weight lifting, and cooking, and I’m attempting to get faster at running.
What I’m Currently Reading
Loading from Goodreads…
What I Read This Year
Loading from Goodreads…
Publications
Papers, preprints, and book chapters, newest first. For a more up-to-date list, see my Google Scholar page.
Do protein representations encode variant function? Diagnosing residue-level protein attributions via embedding-space gradients
Introduces IGLOO, a framework that scores residue-level attributions from protein language model fitness predictors against synthetic landscapes with known per-residue importance.
RA Barghout, S Wu, W Chau, JM Barraza-Chavez, P Driza, R Mahadevan, B Sanchez-Lengeling
kinGEMs: A Robust and Scalable Framework for Resource-Constraint Models through Stochastic Tuning of Deep Learning-Predicted Kinetic Parameters
Integrates deep learning-predicted kcat values from CPI-Pred into enzyme-constrained genome-scale models for metabolic optimization.
RA Barghout, L Chiñas Serrano, B Sanchez-Lengeling, R Mahadevan
Machine Learning for Sequence-to-Function Approaches
Book chapter reviewing machine learning models, benchmarks, and datasets for predicting protein function from sequence.
RA Barghout, M Kirby, A Zheng, L Chinas, M Mohammadi, Z Xu, B Sanchez-Lengeling, R Mahadevan
Graph Data Modeling: Molecules, Proteins, & Chemical Processes
Explores graph-based representations for modeling molecular, protein, and chemical process data.
JM Barraza-Chavez, RA Barghout, R Almada-Monter, A Jinich, R Mahadevan, B Sanchez-Lengeling
CPI-Pred: A deep learning framework for predicting functional parameters of compound-protein interactions
Deep learning approach to predict kinetic parameters from compound-protein interactions.
RA Barghout, Z Xu, J Wu, D Garg, YS Song, R Mahadevan
Advances in generative modeling methods and datasets to design novel enzymes for renewable chemicals and fuels
Review of state-of-the-art generative models and datasets for enzyme design in renewable chemical and fuel production.
RA Barghout, Z Xu, S Betala, R Mahadevan
KinMod database: a tool for investigating metabolic regulation
A database resource for exploring kinetic models and metabolic regulation mechanisms.
K Haddadi, RA Barghout, R Mahadevan