Predicting stress sensitivity of laying hens by identifying genetic, incubation and rearing factors
Location: Hendrix Genetics
Description: It is known that genetic differences exist in stress sensitivity and adaptive capacity, for instance between white and brown hybrids. White hybrids are for instance more flighty and fearful than brown hybrids. Furthermore, also environmental factors can play an important role in determining a flock’s stress sensitivity and adaptive capacity. Here, the parental environment can play a role, but also incubation conditions, the rearing environment, and finally the laying environment. A lot of data is gathered in each of these phases on technical performance, health and welfare, but this data is rarely combined to see whether it can be used to predict stress sensitivity and adaptive capacity of laying hen flocks. The aim of this project is to identify factors in the genetic and rearing background of laying hens that predict stress sensitivity. The starting point will be pattern detection in flock data that is routinely gathered by the breeding company during the different phases of life. Follow-up studies may include more focused data collection or experiments with the aim to develop flock level predictors of adaptive capacity.
Spring 2020 Update
To start with, I had to dig into the literature of the different phases of my project by identifying the stressors and performance traits at each stage and how to connect one phase to the next as shown in fig1.
Furthermore, the data for the different phases was extracted after identifying the data machines (fig2), their functions and which data they store. Some of the data was obtained through direct copying from my colleagues in Hendrix Genetics. The data was then organised by classifying it under the different phases of the project ready to be cleaned. While collecting the data I also learnt a lot by assisting in Hendrix Genetics’ hatcheries, understanding the process flow from eggs to hatched chicks, the regulation and monitoring (fig3) of airflow, CO2, temperature and humidity in the setters and hatchers. Also, the sexing, vaccination & placement of chicks to be transported in the baskets with the temperature recorders. I visited laying hens’ farms and rearing farms (assisting in the delivery of day-old chicks to customers: setting up the rearing area, spraying them against diseases, I must say the chicks are very beautiful and fluffy & I was teaching them how to drink water for the first time, (fig4). It was necessary to get the data and be on the field to understand the processes in each phase of the project, and how to connect the data of all the phases at the end to predict stress sensitivity.
Let’s look at the implementation already done on the first phase (incubation) of this project.
• First, the pre-processing on dataset (2 stressors, 9 break-out features, 125 optimal features).
• Feature Engineering (50 setters, 18 hatchers, 22 lines, 17 farms, 2 generations) with hatchability (target variable).
• Feature Selection with Machine Learning models.
• Exploratory Data Analysis: Hatchability Vs (stressors, Lines, hatchers, generation, Seasons).
• Predicted stress sensitivity between white/brown hybrids using machine learning algorithms (Random Forest, Gradient Boosting Machines) and compared with the traditional linear regression model.
• Analyzed breakout features of incubation & examined variation of stressors in hatchers.
• Presently I am using a clustering algorithm on the genotypes of the white/brown hybrids to cluster them accordingly.
Furthermore, I have submitted my PhD thesis proposal. My secondment to Scotland has been postponed due to COVID-19. I am very happy with the virtual inaugural training school which was supposed to be held in Spain but was well organised in “teams app”. I learnt a lot and I thank everyone for making this happen. I continue to take turns “on teams for now” between Hendrix Genetics, Utrecht University & Wageningen University in collaboration with the ChickenStress European Training Network. I am very grateful for all the support and guidance my supervisors have been giving me, its what keeps me going. Besides working, I enjoy cycling in the Netherlands, you can also see my picture at the office in Hendrix Genetics. I am looking forward to the next step.
Project update 15.10.19
My first week was about settling, getting registered and meeting my colleagues, understanding the organisational structure of Hendrix Genetics, and meeting with my other supervisors in Utrecht University in collaboration with Wageningen University. In addition, I attended training on genetics, participated in meetings, and gathered some resources like thesis and books besides doing some research online about my project. Consequently, the first week gave me a general view of the environment which has enabled me to know who to contact or where to get resources I need for my project.
In my second week, I dived completely into the project, I had a meeting with all my supervisors at the same time, we highlighted the objectives of the project, the scope and an overall outline of the research procedure. I also, visited the hatchery here and discussed with a few people. Considering that I will be working with data from genetics, incubation and rearing factors, I am presently trying to get these data, understand the nature/characteristics of the different datasets and integrate it. This is the first phase of my project and the data part is a very important. Together with my supervisors, we came up with a method of operation and set the dates for our future meetings. Apart from that, I have registered to attend some conferences organised by my university. It has been two weeks since I joined, and I will conclude by saying I have embarked on the journey and I will be taking turns between (Hendrix Genetics and the universities) and collaborating with the ChickenStress European Training Network. I am looking forward to the next steps.