Whole-brain activity in C. elegans
To understand how animals sense external stimuli and environmental cues, we utilize microfluidics to precisely control environmental conditions (e.g., mechanical and temperature stimulus, and multimodal stimuli - mechanical and chemical stimuli) and record neuronal responses in a high-throughput manner.
We have developed integrated microfluidic-based platforms and machine learning algorithms for individualized long-term behavior and longevity tracking to investigate the effects of genetic, environmental (food and temperature levels), and stochastic factors on life- and healthspan in C. elegans.
Individual animals are captured in the chamber
Daphnia magna
We have developed a high-throughput high-content phenotyping platform, a behavioral feature analysis algorithm, and a machine learning-based predictive model to estimate the phenotypic age of Daphnia. With the predictive model, we can estimate phenotypic age under pharmacological perturbations in both long-life and instant assays.
Most of our work heavily relies on the development of microfluidic approaches, which greatly facilitate animal handling and provide unprecedented control of the experimental interventions. Specifically, we have developed a series of microfluidics platforms to deliver controlled external cues such as mechanical, chemical and/or temperature stimuli to a worm while monitoring neuronal responses.
The platforms not only greatly enhance the throughput and robustness of experiments and but also allow to deliver combinations of various stimuli which was not possible using conventional methods. With these platforms, we studied the functional role of sensory and interneurons during developmental and adult animals and identified the neuropeptides that modulate arousal and cross-modal sensitization. Currently, we are developing new microfluidic-based platforms to study age-associated cognitive declines.
Load worms sequentially into the microchannel and deliver mechanical stimuli.
Algorithm for Daphnia phenotyping
Our work relies on quantitative analysis of image-based readouts. To take advantage of images, we have extracted biological meaningful behavioral and morphological features to build a phenotypic profile that is able to describe the healthy state of animals. To do this, we have developed custom image processing algorithms based on both traditional and deep learning approaches.
To predict the healthy state of animals using the extracted phenotypic profiles, we used the machine learning approach. For example, in Daphnia, we built a ML model to estimate animals’ phenotypic age and use it to evaluate the efficacy of various drugs and chemicals.
