1. Home
  2. QTL analysis of multi-environment trials with trait means

QTL analysis of multi-environment trials with trait means

Watch our videos on QTL analysis.

The QTL menus provide easy access to QTL analysis within Genstat. Two sources of data are required: phenotypic data (measurements of trait values) and genotypic/map data (evaluation of genotyping at markers and positions of markers on a genetic map). The description below gives a step-by-step approach to a QTL analysis for a multi-environment trial with trait means. 

Step 1. Put trait means with genotypes and environments (factors) into a Genstat spreadsheet. The spreadsheet should contain data from all trials (environments) stacked on top of each other, with a factor to identify the different trials (environments). You can also include a set of unit errors to take account of unit error carried forward from analyses of the individual environments.
Step 2. Load phenotypic (trait) means and associated factors into the QTL data space using the Stats | QTLs (Linkage/Association) | Data Import/Export | Load Phenotypic data menu item. Select the option for Trait means and load the trait data, genotype factor and environment factor (and unit error if present). The QTL data space Stats | QTLs (Linkage/Association) | View QTL Data Space will now show these variables under the Phenotypic means folder.
Step 3. Run genotype by environment exploratory analysis using the menus under the Stats | QTLs (Linkage/Association) | Phenotypic Analysis menu item. These can be used to investigate the distribution of data across environments (Summary statistics) or to investigate genotype-by-environment interactions (AMMI & GGE biplot).
Step 4. Select model for covariation across environments using the select best variance-covariance model menu opened using the Stats | QTLs (Linkage/Association) | Phenotypic Analysis | Select Best Variance-covariance model for Multiple Environments menu item. This menu allows comparison of a set of across-environment variance-covariance models and the best model is selected using AIC or SIC (see Options).

Using the select best variance-covariance model menu:

  • Unit errors generated from the preliminary single trial analyses can be incorporated into the estimation process using the Options menu. Use of these errors allows the relative precision of data from different trials to be incorporated in the analysis.
Step 5. Load genetic marker and map data using the Stats | QTLs (Linkage/Association) | Data Import/Export | Load Genotypic (Marker and Map) data menu item. Choose the file type that contains the data (Flapjack text genotype/map, R/QTL csvs/csvsr and MapQTL loc/map file formats) and load the data. The QTL data space will now show these data structures under the Genotypic data folder.
Step 6. Check marker and map data using the display genetic map and genotype data plots menus (opened by selecting the Stats | QTLs (Linkage/Association) | Data Exploration | Genotypic Data | Display Genetic Map and Stats | QTLs (Linkage/Association) | Data Exploration | Genotypic Data | Genotype Data Plots respectively).
Step 7. Calculate genetic predictors for use in QTL interval mapping using the calculate genetic predictors menu opened using the Stats | QTLs (Linkage/Association) | Genotypic Analysis | Calculate Genetic Predictors menu item. The QTL data space will now show these data structures under Genetic predictors folder. Using the calculate genetic predictors menu:

  • This calculation may take some time to run. Do not attempt to run further analysis until it has finished (ie. when Genstat histogram in Taskbar goes back to green)
Step 8. Run simple interval mapping using the Initial scan button on the Single Trait Linkage Analysis (Multiple Environments) menu (opened using the Stats | QTLs (Linkage/Association) | QTL Analysis | Single Trait Linkage Analysis (Multiple Environments) menu item). This does an initial scan for candidate QTLs as main effects or QTLxE interactions, which are saved for further analysis. By default, the results of the scan are plotted and the candidates are displayed in the output.

Using the Single Trait Linkage Analysis (Multiple Environments) menu:

  • This interval mapping may take some time to run. Do not attempt to run further analysis until it has finished (ie. when Genstat histogram in Taskbar goes back to green)
  • Unit errors generated from the preliminary single trial analyses can be incorporated into the estimation process using the Options menu. Use of these errors allows the relative precision of data from different trials to be incorporated in the analysis.
Step 9. Run composite interval mapping using the Scan with cofactors button on the Single Trait Linkage Analysis (Multiple Environments) menu. By default, the cofactors are set as the candidate QTLs from the previous scan. This does a scan for additional candidate QTL (as main effects or QTLxE interactions) in the presence of the cofactors, although cofactors within a set window (see Options) are ignored to avoid problems of collinearity. A modified set of candidate QTLs is saved. By default, the results of the scan are plotted and the candidates are displayed in the output.

Using the Single Trait Linkage Analysis (Multiple Environments) menu:

  • This interval mapping may take some time to run. Do not attempt to run further analysis until it has finished (ie. when Genstat histogram in Taskbar goes back to green)
  • Unit errors generated from the preliminary single trial analyses can be incorporated into the estimation process using the Options menu. Use of these errors allows the relative precision of data from different trials to be incorporated in the analysis.
Step 10. Run composite interval mapping final model selection using the Select final QTL model button on the Single Trait Linkage Analysis (Multiple Environments) menu. By default, the full set of candidate QTLs (as main effects and QTLxE interactions) are the set selected from the previous scan. The default first step is backwards selection from this candidate step, followed by estimation of the QTL effects for all QTLs (as main effects or QTLxE interactions) retained in the selected model.

Using the select final model (multi-environment) dialog:

  • Unit errors generated from the preliminary single trial analyses can be incorporated into the estimation process using the Options menu. Use of these errors allows the relative precision of data from different trials to be incorporated in the analysis.

See also

Updated on April 24, 2019

Was this article helpful?