Predicting new disease phenotypes from cow milk mid infra-red spectral data using Deep Learning

Mid infrared (MIR) spectroscopy of milk samples is an internationally used non-invasive method for the prediction of fat and protein during routine milk recording. It has been successfully and widely demonstrated that MIR spectroscopy of milk samples can be used to predict an increasing number of different phenotypes in dairy cows due to the chemical imprint left by biological processes in the milk 1,2. This method of prediction is increasingly valuable as an efficient and effective low cost tool for rapid measurement of expensive and often difficult-to-record phenotypes 3. Currently, the applicants hold complete spectral profiles for over 1.4 million cows from over 4,000 national herds; data are stored and transferred to SRUC nightly. Calibrating spectral data using phenotypes by PLS regression has delivered several successful quantitative analysis tools (e.g., body energy), however, this requires data be measured on a continuous scale and normally distributed. In the case of discrete data (e.g., categorical, binary, etc.) the usual methods for developing prediction equations have proved inefficient and result in lower accuracy. As such there is a requirement for alternative and novel mathematical/statistical techniques to better utilise MIR spectra; a requirement that can be met using machine learning. Deep learning, a branch of machine learning, employs algorithms and techniques that are better able to make use of the increasingly huge datasets and advances in computer technology of the present day. This project will use machine learning techniques to develop robust predictive models of two key disease traits using MIR data, and develop more accurate prediction equations for national phenotypes. The project will provide training in a number of highly valuable areas inc. very large data handling, Deep (machine) Learning computing methodologies, data driven biology, application of novel techniques to real world problems and finally and possibly the most valuable, engagement with a commercial company applying Deep Learning in the field.

Applications:
Completed application form along with your curriculum vitae should be sent to our PGR student team at [Email Address Removed]

References:
(Two required. Please inform your referees of this and note that it is your responsibility to ensure that references are provided by the specified deadline)

Downloads:
Application form –https://www.ed.ac.uk/files/atoms/files/eastbio_application_form_july_2018.doc.
Reference request form – https://www.edweb.ed.ac.uk/files/atoms/files/eastbio_reference_request_form_july_2018.doc