Faculty Profile: Amit Singer creates algorithms and reconstructs 3D images of molecules

Monday, Nov 29, 2021
by Sharon Adarlo

Amit Singer, the Herbert E. Jones, Jr. '43 University Professor of Mathematics, and Professor of Mathematics and the Program in Applied and Computational Mathematics

Background:

Singer has been a member of Princeton University’s mathematics department and Program in Applied and Computational Mathematics since 2008. He is also a member of the executive committee at the Center for Statistics and Machine Learning (CSML).

Singer majored in physics and mathematics at Tel Aviv University, where he earned a bachelor’s degree in 1997. From the same university, he earned a doctoral degree in applied mathematics in 2005. His thesis, titled “Diffusion Theory of Ion Permeation through Protein Channels of Biological Membranes,” won the Nessyahu Prize for Best Ph.D. Thesis in Mathematics in Israel.

In his academic career before joining Princeton, Singer pursued studies at Rush University Medical Center in Chicago during his graduate years. He was also a Gibbs Assistant Professor in applied mathematics at Yale University. During that time, he was a visitor at the Institute of Pure & Applied Mathematics at the University of California, Los Angeles.

In addition to numerous grants from bodies such as National Science Foundation and DARPA, he has been honored with a Sloan Research Fellowship, Presidential Early Career Award for Scientists and Engineers, and the National Finalist for Blavatnik Awards for Young Scientists, among other awards.

He has been listed as co-author on more than 100 papers during his academic career and has been associate editor for publications such as SIAM Journal on Mathematics of Data Science, Applied and Computational Harmonic Analysis, and several others.

Research:

Singer’s research is divided into two main components: one - developing algorithms and advanced computational methods to analyze data, and two – using these tools to determine the three-dimensional structures of molecules.

“The first component is more theoretical and foundational. I develop algorithms and do mathematical analysis of existing and new algorithms to analyze data. This data can be quite complex with very high dimensionality. This is data with a large amount of noise or the data set can be quite large. So, my lab members and I come up with solutions or tools to process and analyze this data. Such contextualization programing is one part of my research,” he said.

The second part, the modeling of molecules, is the application of these data science tools, Singer said. The modeling of these molecules starts with cryo-electron microscopy or cryo-EM, an emerging technique to take images of molecules. The process consists of placing molecules such as proteins in a liquid solution and then freezing these samples to -180 °C. Electrons are then shot at the sample and then a camera captures the electrons in order to make an image. Jacques Dubochet, Joachim Frank and Richard Henderson, the scientists who developed this process, received the Nobel Prize in Chemistry for the technique in 2017.

Singer said this technique is a great alternative to existing methods such as crystallography which involves growing a crystal structure of molecules. Unfortunately, proteins are hard to crystalize. That’s why scientists have become interested in cryo-EM because it does not require crystallization, said Singer.

Singer takes these cryo-EM images, which can be quite noisy and are in 2D, and uses the computational tools he and his lab have developed to reconstruct the 3D structure of these molecules that can be ultimately refined to the atomic level. Going from 2D to 3D is a challenging endeavor because the images are very noisy, the orientation of the proteins are unknown, and they can flex or bend, Singer said.

A big project he and his team have been working on is a software package that can process and analyze these cryo-EM images with little human intervention.

Algorithms for Single Particle Reconstruction or ASPIRE offers fast processing of data while producing accurate 3D images of proteins and other molecules. Vineet Bansal, senior software engineer jointly appointed to CSML and the Princeton Institute for Computational Science and Engineering (PICSciE), was instrumental in helping Singer’s lab in initially developing the software package. More on his role can be read here.

“Our team’s long-term goal is to develop a software package to be used for drug discovery and research and by other cryo-EM and image processing developers,” said Singer.

ASPIRE, which is open source, can be accessed at this website.