Back to CassavaBase Manual |

Creating prediction models for multiple traits and predicting breeding values for multiple traits in a training population

Prediction models are fit per trait; it is a univariate modeling. However, you can simultaneously fit models for multiple traits from a single training population. To do so:

1. Go to the training population page of your interest, which may be made up of individuals from a single trial or combined trials or a custom list of individuals. Once on the population page, from the list of traits select any number of traits and click 'Build model'.


2. After a few minutes of wait, you will get a multiple models output page (see below). Click a trait name and you will get details of the model output for that trait. See here for an example of single model output detail.


Estimating breeding values of selection candidates

To simultaneously predict breeding values of selection candidates for all the traits with prediction models, click on the link of your selection population. Once prediction is complete, the 'predict' link will be replaced with the traits names. If the list of traits is missing any of the traits you have fit prediction model, first check if the models accuracy is greater than zero, if it is so then click the selection population again. To see details of the prediction output of a trait, click on the respective trait name.


Estimating genetic correlation

To estimate genetic correlation among traits you have estimated breeding values for, from 'Genetic correlation analysis' section select the population for which you have estimated breeding values and click 'Run correlation'. You will get an output something like below.


Calculating selection index

To calculate selection indices for individuals based on their breeding values for multiple traits, from the 'Calculate selection index' section select the population of your interest, fill in the relative weights of the traits and click 'calculate'. You will get a table of indices as well us a heatmap of correlation among the traits' breeding values and selection indices.


Back to CassavaBase Manual |