Mathematics and Natural Sciences

Laura Finzi, PhD

PROFESSOR, ECAS, PHYSICS

Single molecule frameworks for studying transcriptional interference

I am applying for funding to support research that leverages a new microscope acquired with a recent NSF award, the Correlative Trap (C-Trap) microscope. The proposal is articulated in three Aims. The first Aim is to develop two molecular frameworks that could have broad applications in the study of DNA processing motors, including transcribing RNA polymerases (the enzymes responsible for copying the DNA genome into RNA during the process known as transcription). The second Aim is to obtain preliminary data about the interference with a transcribing RNA polymerase by proteins that normally decorate genomic DNA. The third Aim is to obtain preliminary results about the capacity of RNA polymerase to move past defects in genomic DNA, such as mis-incorporated ribonucleotides, which are subunits of RNA, not DNA.
The molecular constructions developed in Aim 1 will be the foundation of both Aims 2 and 3. The results of Aim 2 will be assembled into a new application to the NIH-National Institute of General Medical Sciences (NIGMS), while those of Aim 3 will be packaged for a new application to the NSF-Molecular and Cell Biology (MCB) Cluster. Given the fundamental significance of these two Aims for the understanding of the molecular mechanisms of transcription by RNA polymerase and that NSF welcomes applications that leverage instruments acquired with the Foundation support, I am confident about obtaining extramural funding for our innovative research program.

Debjani Sihi, PhD

ASSISTANT PROFESSOR, ECAS, ENVIRONMENTAL SCIENCES

Building Soil Organic Matter through Biochar Amendment: A Climate-Smart Approach to Ensure Food Security

Soil organic matter (SOM) regulates the capacity of soil to provide regulatory ecosystem services. However, measuring changes of SOM effectively warrants taking its differences into account. Current efforts (climate-smart land-management practices) to promote SOM storage miss a key point. An improved understanding of how different SOM fractions (particulate organic matter, POM and mineral-associated organic matter, MAOM) work and how our lifestyle choices change their relative distribution is of paramount importance to meet the two most pressing problems facing our planet: climate change and food security.

We can think about POM and MAOM fractions as the amount of money in bank, where POM can be a proxy for the checking account and MAOM represents the savings account. While the checking account generally receives more money from each paycheck, we also spend that money quickly to cover our routine expenses. However, the savings account receives less money per payday, but that money remained saved for a long-time. Likewise, when we grow food using heavy machines for plowing, it causes rapid loss of POM than MAOM. Because of their very different lifetimes, measuring POM and MAOM have important environmental and socio-economic implications.

This study will evaluate the dynamics of POM and MAOM under biochar amendment. We choose biochar because this is a climate-smart approach, which holds promise for a sustainable food future in nutrient-poor soils of tropics and subtropics. We expect that the findings will provide an innovative solution to evaluate the impact of climate-smart land-management practices, which have important policy implications.

Li Xiong, PhD

PROFESSOR, ECAS, MATHEMATICS AND COMPUTER SCIENCE

PREMED: Privacy-Preserving, Robust, and Efficient Computational Phenotyping using Multisite EHR Data

Computational phenotyping is the process that distills accurate and concise clinical concepts, or phenotypes, such as disease subtypes, from the Electronic Health Records (EHR) data. It is essential for a variety of purposes including population management and observational and interventional research. An open challenge is how to harness the data across multiple institutions for more effective phenotyping while preserving patient privacy at each site.
We propose a federated framework for Privacy-preserving, Robust, and Efficient computational phenotyping using Multisite EHR Data (PREMED) while keeping data within local sites. Our preliminary studies have showed that it is possible for federated phenotyping to achieve comparable accuracy to centralized model with integrated data. However, challenges remain in order to achieve communication efficiency, rigorous privacy, and robustness against potential failures of local sites. Our project addresses the three interrelated challenges with a holistic approach. Our hypothesis is that communication-efficient techniques, such as compression of intermediate results from local sites, have intrinsic benefit to privacy and robustness due to the compressed and hence obfuscated communication. For this URC project, we will develop communication-efficient techniques and investigate their intrinsic privacy and robustness benefit. In collaboration with Dr. Siva Bhavaniat the School of Medicine, we will conduct case studies using EHR data on Sepsis from multiple hospitals. The preliminary results will enable us to seek external funding to further develop explicit private and robust methods that exploit the implicit benefit of communication-efficient techniques; and conduct larger-scale and broader case studies with multiple institutions and other conditions.