Inference and prediction of GxExM and genetic gain from TRICOTS variety trials

Submitted by hugo.dorado on
Organizational Context
Name
Hugo Andres Dorado Betancourt
Chairgroup
Plant Production System (PPS)
Graduate school
PE&RC
Start date of project
Abstract

This research evaluates and develops methodologies for the estimation of genetic value and genetic gain for on-farm trials. In the first stage, to test the methodologies, the yield and other response variables of several cultivars are emulated via Montecarlo simulation, this dataset will be randomly generated based on probabilistic distributions with contrasting parameters for each variety using R. In the second stage, the developed methodologies will be used to analyze real on-farm trials records. This data will be collected by Alliance Bioversity-CIAT using the TRICOTs methodology (https://alliancebioversityciat.org/tools-innovations/tricot-triadic-comparisons-technologies). Permission for the use and publication of the TRICOT data will be handled by the same institution. This research is part of the 1000farms project which aims to accelerate the breeding of new climate-adapted and farmer-preferred crop varieties (https://1000farms.net/resources/). The project will collect data in some regions of Africa that will be defined through the project: Benin, Burkina, Ethiopia, Ghana, Kenya, Malawi, Mali, Mozambique, Nigeria, Rwanda, South Africa, Tanzania, Uganda, Zambia, and Zimbabwe.

Who's collecting the data

Part of the data will be generated via simulation of probabilistic distributions using R. The data will be generated by Hugo Dorado (pH Student), Joost van Heerwaarden (Supervisor), and students working at the project 1000 Farms. Moreover, we will use data from small-scale variety trials in different regions of Africa, on-farm data collection will be in charge of Alliance Bioverstity CIAT, which is the organization that leads the project.

Who's analysing the data

The data will be analyzed by Hugo Adres Dorado pHD Student in WUR, Joost van Heerwaarden professor in WUR, and students working at the project 1000 Farms in WUR.

Location short term storage

All data will be stored on my local harddisk in a folder called Thesis.

Within this Thesis folder, I'll create per chapter the folders: Data, Model, Paper and Scripts. The Data folder has two sub-folders called: Raw and Processed.

Folder contents:

  • Data - Raw sub-folder: Contains all raw data and meta-data (a description of your data).
  • Data - Processed sub-folder: Contains all processed data. 
  • Model folder: Complete listing of the model and the model results & analysis.
  • Paper folder: Text of a chapter / paper.  
  • Scripts folder: Contains all scripts used.
Backup procedure

The complete content of my local Thesis folder will be stored on my YoDa-drive. 

During periods I'm abroad, I'll backup the complete content of my local Thesis folder to a Dropbox or MS-Onedrive Thesis folder and share the contents with my supervisor(s). 

Research data with value for long term storage

Scripts and dataset simulated

Research data excluded for long term storage. Why?

No data will be excluded

Plans for sharing data?

We plan to publish scripts and data in open repositories

How to access data once you leave?

The code developed for data analysis will be uploaded in a GitHub repository, which will be open to anyone. The data will be published at the Alliance Bioversity CIAT datavese which ensures that will be well documented and accessible.

Specific funders requirements for sharing data, or to impose embargo?

No requirements from funders

Other parties involved? Agreements on data sharing?

Data on variety trials will be collected by Alliance Bioversity CIAT, they will publish this dataset under his own policy of  data sharing