Eviatar Yemini

Research Overview

Eviatar Yemini focuses on the development and evolution of the nervous system at the resolution of individual neurons, with a primary emphasis on the nematode C. elegans. His work integrates computer science, behavioral genetics, and cellular neurobiology to investigate how neural circuits reconfigure during development to meet the behavioral requirements of different life stages. By developing high-throughput behavioral tracking systems and in vivo cell-type identification tools, Yemini aims to establish a systems-biology model of the whole nervous system that connects genetic and molecular changes to cellular activity and organismal behavior.

Research Trajectory

Yemini’s research trajectory is characterized by a transition from high-throughput behavioral quantification to the development of sophisticated molecular and computational tools for whole-brain imaging. His early work was rooted in the need for automated, quantitative methods to analyze animal behavior at scale. This led to the development of the Multiworm Tracker (MWT), a system designed to facilitate genetic and drug screens by transforming raw movement into quantitative data. This period of his career focused heavily on the intersection of computer vision and behavioral genetics, establishing a foundation for large-scale phenotyping.

Following the establishment of these behavioral tools, Yemini’s focus shifted toward identifying the underlying neural architectures responsible for the observed behaviors. A significant challenge in C. elegans research was the difficulty of identifying specific neurons in live, behaving animals. To address this, Yemini developed "NeuroPAL" (Neuronal Polychromatic Atlas of Landmarks), a method utilizing color barcodes for in vivo cell-type labeling. This tool allowed researchers to identify every neuron type across the entire nervous system of the worm, facilitating the mapping of neural communication and the visualization of dynamic changes in cell identity.

In recent years, Yemini has expanded his research to include transcriptomics and functional connectomics. As a principal investigator, he has overseen projects that integrate bulk and single-cell RNA sequencing to refine transcriptomic profiles of individual neurons. His current work also explores the temporal evolution of neural circuits, particularly how they change during juvenile and sexual development. This involves applying Iterative Matrix Normalization for Efficient Second Order Deep Learning and other computational frameworks to analyze multi-dimensional neural activity and position.

Current Directions

The Yemini Lab currently pursues several interrelated research themes aimed at understanding the logic of nervous system reconfiguration. One primary area of focus is the study of juvenile and sexual development, specifically how the nervous system grows and changes to accommodate the shifting behavioral needs of the organism. This involves using NeuroPAL to track cell fate and identity across different developmental stages, providing a high-resolution view of how transcription factors and other genetic regulators influence neural architecture.

Another major direction is the mapping of neural communication. By combining whole-brain imaging with the cell-type identification capabilities of NeuroPAL, the lab seeks to map communication patterns at single-cell resolution across the entire nervous system. This work extends to the study of chemosensory responses and the determinants of neuron position, often involving comparative studies between different sexes of C. elegans.

Furthermore, Yemini is actively involved in expanding the NeuroPAL methodology to other model organisms. This effort aims to provide the broader neuroscience community with tools for in vivo cell identification in more complex systems. The lab also continues to develop computational frameworks for analyzing neural activity, such as multi-dimensional hyperbolic embedding and unsupervised learning approaches to reveal dynamic functional connectomes.

Representative Work

The development of high-throughput behavioral analysis was a foundational component of Yemini's early research. In Tracking Movement Behavior of Multiple Worms on Food, Yemini introduced the Multiworm Tracker (MWT), a system capable of simultaneously monitoring dozens of nematodes. The MWT utilized real-time image processing to quantify movement while filtering out complex "touching" events that typically confound automated trackers. This work enabled long-term, unattended experiments, providing the quantitative rigor necessary for large-scale genetic and pharmacological screens.

The utility of these tracking tools is evidenced by Yemini's extensive contributions to the C. elegans behavioral database. A series of experimental records, such as those for the strains Aq1422, Rb559, and My16, document the behavioral phenotypes of various mutants and wild isolates. These entries, which include data on strains like Aq2933, Vc2324, Mt1179, Qz81, Rb1911, Rb919, and N2, represent a systematic effort to map the genetic determinants of movement. By standardizing the collection of behavioral data across hundreds of genotypes, this work provided a resource for the systems-level analysis of how specific genes influence organismal output.

More recently, Yemini has participated in the development of computational methods to interpret the complex data generated by whole-brain imaging. In Learning dynamic representations of the functional connectome in neurobiological networks, he contributed to an unsupervised approach for learning dynamic affinities between neurons in behaving animals. This method employs non-negative tensor factorization (NTF) to organize pairwise non-linear affinities from calcium activity traces, revealing how different neuronal communities form and dissolve over time. This work addresses the contrast between static synaptic connectivity and the dynamic nature of neural function, providing a framework for understanding how various combinations of neurons cooperate to effect specific behaviors.