Speakers: Mihaela Sardiu, PhD

Mihaela Sardiu, PhD

Associate Professor
Biostatistics & Data Science
University of Kansas Medical Center


Mihaela Sardiu has a solid grasp in multiple disciplines with a broad background in physics and computational biology with specific training in protein folding theory and statistics. She started  graduate study in theoretical physics at Florida Atlantic University and then continued to graduate study at the National Center for Biotechnology Information at the National Library of Medicine at the National Institutes of the Health under the supervision of Dr. Yi-Kuo. She showed a solid connection between the random matrix theory (one of the most important mathematical advances in theoretical physics) and the score statistics of sequence alignment. This work was published in Physics Review E (2005). During my Ph.D. She introduced the “target-focusing” concept, in which an addition of nonspecific interactions from evolutionarily selected contact pairs will help large proteins to fold more efficiently. This revolutionized the old consensus that proteins fold fastest in a completely downward energy landscape. This work was published in Biophysical Journal.

Mihaela Sardiu worked with Dr. Michael Washburn at the Stowers Institute for Medical Research as a postdoctoral, where she continued to apply physics and statistics skills in the field of dynamic networking with focus on protein interaction networks. At the Stowers Institute, she has been investigating how to best utilize the quantitative information generated by advanced protein mass spectrometers to advance out understanding of protein complexes and protein interaction networks.  She developed the first probabilistic approach for assembling protein interaction networks based on label free quantitative proteomics.  This work was published in the Proceedings of the National Academy of Sciences of the United States of America. In addition, she developed approaches for determining protein complex architectures using quantitative proteomics and complex disruption.  More recently, she utilized Topological Data Analysis for the computational analysis of protein interaction network datasets.  These studies have led to the development of an approach to score protein interaction networks in such a way as to predict direct protein-protein interactions in a dataset.  This work was recently published in Nature Methods. She successfully collaborated with other researchers and published results in high-impact peer-reviewed papers.

Identifying Robust Structures in Multiscale Omics’ Data

With large enough data significant patterns can be detected. For example, protein-protein interaction networks can detect groups of proteins that form protein complexes or large functional networks. The analysis of protein complexes and interaction networks, and their dynamic behavior as a function of time and cell state, are of central importance in biological research. Likewise, RNA-sequences, Chip-seq or flow cytometry data sets can be clustered coarsely to identify broad categories of functional genes/proteins or specific cell subtypes. While several approaches have being developed to detect structures in biological data, these methods were optimized for very specific types of data and are not generally applicable in many experiments. To this end, we have recently demonstrated the great utility of the topological score (i.e. TopS) to track the hierarchical arrangement of the macromolecular complexes and to detect rare events in omics data. Building upon this work, we sought to integrate topological score (i.e. TopS) with existing algorithms for network structure detection, resulting in a practical multiscale approach for many biological networks.