Abstract:

In the demesne of neuropsychiatric drug discovery, success has been scarce and preferably serendipitous. An original scientific reason is the relationship between genetic target and neurological function is not well understood. And a principle technical constriction for drug discovery is a high throughput in vivo screen with highly accurate and sensitive readout in animal model has been lacking. Our goal is to explore the potential of zebrafish as an in vivo drug discovery model for behaviors and neurological functions, which are specifically believed relevant to neuropsychiatric disorders, such as autism spectrum disorder (ASD) and schizophrenia. To illustrate the relationship between genetic mutation and behavioral changes in an animal model – the zebrafish, CRISPR genome editing are applied to disorder brain functions, and subsequently animal behavioral changes in multivariate dimensions are measured by assorted Machine Learning algorithms, such as Support Vector Machines (SVM), and Deep Learning. We demonstrate Machine Learning algorithms are powerful tools for rational feature inference and discovering subtle solitary or social behavioral changes, which reflects similar aspects of schizophrenia and ASD phenotypes. Furthermore, by deploying High Performance Clusters (HPC) and Robotic Automated Screens (RAS), diverse behavioral conditions can be investigated for large number of animals at high speed. The hope is that over the long term these may serve as systems relevant to in vivo psychiatric drug discovery in combination with high spatiotemporal in vivo multiphoton imaging using the zebrafish model.