A high-throughput imaging and classification system for fruit flies

Flysorter, Llc, Seattle WA

Investigators

Linked publications & trials

Abstract

PROJECT SUMMARY / ABSTRACT In this Phase I SBIR application, FlySorter proposes to development a high throughput imaging and classification system to aid research with fruit flies, a widely-used model organism relevant to both basic science as well as studies in human health. The use of animal model systems is essential for research in almost all aspects of biology: genetics, development, neuroscience, disease, physiology, and beyond. The fruit fly ? Drosophila melanogaster ? is small and easy to care for, but is complex enough an organism to provide a wealth of information that directly relates to human biology and health. Over 75% of human diseases with a genetic basis (including depression, alcoholism, certain forms of cancer, and many more) are either present or have an analog in Drosophila. Modern genetic tools, such as CRISPR/cas9, allow the creation of transgenic flies that provide the opportunity to study diseases, pathways and systems that don?t exist naturally in Drosophila. With these advances, fruit flies are becoming more frequently subjects for drugs screens. For all the advances in the biological tools and techniques applicable to flies, however, the limiting factor in many experiments is the manual labor involved in a few common tasks: moving flies from vial to vial or other lab equipment; classifying and sorting flies by sex, eye color and other phenotypes; and collecting virgin female flies before they mate so that they can be used in controlled crosses, etc. FlySorter?s patent-pending fly dispensing mechanism can reliably deliver a single organism from a vial containing hundreds of awake flies, and our novel FlyPlate system allows storage of individual flies in custom 96 well plates. FlySorter?s robotic fly handling system, co-developed with the de Bivort Lab at Harvard, is capable of manipulating and transporting those individual flies between vial, 96 well plate, and experimental apparatus. The next piece of the automation puzzle to solve is high throughput imaging and classification. To accomplish this goal, FlySorter will: 1) complete a prototype automated image capture hardware system; 2) adapt state-of-the-art computer vision and machine learning algorithms for use on Drosophila; and 3) build a module that can physically sort the classified flies into different vials. Once integrated into the existing FlySorter product ecosystem, this imaging and classification module will greatly expand the kinds of experiments and screens that can be automated, allowing for the study of larger populations or a wider variety of flies, reducing the impact of human error, and freeing up valuable time for researchers.