Skip to main content
ARS Home » Southeast Area » Stoneville, Mississippi » Genomics and Bioinformatics Research » Research » Research Project #436241

Research Project: Sex Determination of Eggs by High-speed Volatile Compound Mass Spectrometry and Machine Learning

Location: Genomics and Bioinformatics Research

Project Number: 6066-21310-005-40-T
Project Type: Trust Fund Cooperative Agreement

Start Date: Aug 1, 2019
End Date: Aug 1, 2021

Objective:
In-ovo sexing technology has the potential to eliminate male chick culling in the poultry industry. This proposal focuses on the novel application of volatile chemical detection to rapidly determine the sex of eggs prior to incubation without pretreating the shell. Volatile chemical compounds have been used to successfully determine the sex of quail eggs at day 1. We generated pilot volatiles data for this proposal from 20 intact chicken eggs, and determined the sex of the chicks in those eggs. These data show clear differences in sex at day 2-3. Volatiles collection is simple, non-invasive and sensitive, but the methods has been largely overlooked since the traditional collection of volatiles data requires gas chromatography/mass spectrometry which is time consuming. The work in this proposal will verify the utility of volatiles data for sexing eggs and bring it to production speed by, identifying key compounds, designing machine learning classifiers, creating production-compatible egg trays for collecting volatile gases in thousands of eggs in parallel, and developing a gas chromatography-free method to detect volatiles at a rate 2,000 samples per hour using proton transfer reaction a mass spectrometer that samples automatically from volatile collection egg trays.

Approach:
Moving volatiles technology from idea to prototype in 24 months requires several key steps: Month 1-3: the collection of volatiles profiles from a range of breeds, and production facilities using conventional Gas Chromatography/Mass Spectrometry (GC-MS). Month 2-4: the selection of informative features (i.e. volatile compounds) for machine learning classifiers Month 3-6: the development of a proton transfer reaction mass spectrometry (PTR-MS) protocol to detect a specific suite of volatile compounds in a 100-200ms response time. Month 6-12: Contacting with Prototyping firm on design of trays for headspace collection. Month 6-12: Development of sampling system to collect gas from headspace and transfer to PTR-MS Month 9-15: writing interface software to connect the PTR-MS to a control computer and make machine learning (ML) classification in real time Month 15-18: Large scale collection of volatiles data from multiple breeds and producers to improve classification. Month 18-21: preparation of patent application and manuscript, interactions with potential licensees (Damtech, Sanovo, Avitech, etc).