CAREER Data Driven Approaches for Investigating Olfactory System Heterogeneity

CAREER Data Driven Approaches for Investigating Olfactory System Heterogeneity CAREER: Data-Driven Approaches for Investigating Olfactory System Heterogeneity Dr. Gerkin will be writing customized data acquisition, analysis, and logging software for the Castro lab that will allow for semi-automated physiological phenotyping. In Dr. Castros original grant, this was described as locally crowd-sourced physiology, owing to the fact that the data will be collected by a large group of experimentalists, each with relatively modest amounts of time to invest in data collection (college students). The front end of the data acquisition suite will be a minimalist GUI written in Igor Pro. This will run on a dedicated computer in an electrophysiology rig in the Castro lab using redundancy measures as outlined in the data management plan. The acquisition suite will control an ITC18 (Instrutech) digitizer through appropriate Igor XOPs, and will be extendable to accommodate other A/D devices. The acquisition software will accommodate pushbutton physiology: once a cell is patched, its health and seal quality will be automatically and intermittently monitored via responses to test pulses in current clamp and voltage clamp. In early iterations, the investigators may train a cell quality classifier on labeled instances of good vs. bad seals, and healthy vs. unhealthy cells. Each recorded cell will be subjected to a battery of stimulus pulses, including square pulses, sinusoidal chirp pulses, and noise injection. The stimulus library will encompass and extend the stimulus set employed by the Allen Brain Institute in their neuronal cell-typing project. As the data are collected, there will be a variety of readouts and graphical displays that relate ongoing data collection to the existing library of previously analysed cells. The acquisition GUI will include standard e-notebooks for experiment logging, and metadata tagging. After data are collected, they will be automatically analyzed, and physiological parameters extracted. An experiment summary will be auto-generated for each cell and saved as a PDF. Raw data will be saved in a relational database, and each cell will also be represented as a feature vector that will populate the columns of a data matrix. A web application for reviewing and reporting on the collected data will also be developed. Basis for selection of the subrecipient organization Dr. Castro and Dr. Gerkin are long-time collaborators with a strong rapport, overlapping in adjacent labs in grad school, and also overlapping briefly as postdocs in Nathan Urbans lab. Dr. Gerkin is intimately familiar with Dr. Castros research goals, which will greatly facilitate (and save expenses on) the software design process. Additionally, Gerkin and Castro intend to make high-throughput cellular phenotyping (and hence this software) a central part of a future collaboration. Having this software initially written by Dr. Gerkin will therefore make it much more readily customizable, flexible, and extensible in the future. Finally, from queries of high quality software vendors and custom software designers, Dr. Castro was informally quoted figures several times higher than the agreed-on price described here.