Project:	Developing inversion methods for a bioisotopic metabolic model
Disciplines:	Computer Science, Environmental Science and Engineering
Mentor:	Alex Sessions, Professor of Geobiology, (GPS), sessions@caltech.edu
Mentor URL:	https://sessions.caltech.edu  (opens in new window)
AO Contact:	Yeonsoo Park, ypark2@caltech.edu
Background:	The carbon isotopic compositions of organic molecules have long been measured to gain insight into natural processes. However, it has been difficult to interpret such signals quantitatively in terms of the biochemical (metabolic) processes that create the molecules. We recently developed software (‘QIRN’) to build and run a forward bioisotopic model that takes information on the system of interest (initial conditions, reaction network structure, and associated isotope properties) to predict the isotopic composition of compounds in the network. By comparing measurements to the software’s predictions, we can infer characteristics of the studied system, from input sources to the relative contribution of different processes within the system and more. Application of this tool to complex metabolic networks (e.g. carbon metabolism by microorganisms) is limited by the number of unknown properties of the network – in particular isotope effects – that grows with the size of the network being modeled. Hence, we are developing an inverse modeling approach that uses known isotopic compositions of measurable compounds to predict the unknown properties of the network that are difficult to learn experimentally.
Description:	The student will work with their mentor to develop code to invert QIRN, solving for unknowns in the model through parameter optimization. The mentor will provide the necessary background in isotope biogeochemistry and the basics of the software’s code written in Python. Together, we will brainstorm ways to approach the optimization problem, including a literature review of relevant techniques. Then, the student will implement the discussed method(s) in the software code. Subsequently, the code will be tested by comparing the obtained predictions to experimental data (provided by the mentor). If time permits, the student will also seek ways to improve the computational efficiency of the code, and/or user interface to enable network visualization.
References:	- https://web.gps.caltech.edu/~als/ - Mueller, E. P., Wu, F., & Sessions, A. L. (2022). Quantifying Isotopologue Reaction Networks (QIRN): A modelling tool for predicting stable isotope fractionations in complex networks. Chemical Geology, 610, 121098. (https://www.sciencedirect.com/science/article/pii/S0009254122003928)
Student Requirements:	The student should have experience in coding with Python (those with experience with understanding and building from pre-existing Python codes will be preferred). A basic understanding of biochemistry and metabolism will be helpful but is not required.
Programs:	This AO can be done under the following programs:   Program    Available To        Amgen Scholars    Non-Caltech students only          SURF    Caltech students only  Click on a program name for program info and application requirements.