breedtools: design, analysis and 'Breedbase' storage of plant breeding trials

breedtools designs and analyses plant breeding field trials, from early generations with a single plot per entry to fully replicated trials, and saves the designs into a 'Breedbase' database. The tutorial, vignette("breedtools"), walks through every design and analysis.

Which design?

Start with design_options() when the number of entries is known but the field dimensions, checks or replication are still to be decided.

One plot per entry, one location: augmented row-column design with replicated checks.
One plot per entry per location, several locations: augmented partially replicated (p-rep) design, without checks.
Every entry in every replicate: resolvable (spatial) row-column design.

Augmented row-column designs

Unreplicated entries plus replicated checks in rows, columns and blocks (Federer, Nair and Raghavarao, 1975; Piepho and Williams, 2016), with equal or unequal row and column groups, complete or incomplete blocks of checks, and Latin or D-optimal check placement that keeps checks apart.

Plan: augmented_row_column_layouts(), augmented_row_column_capacity()
Design: augmented_row_column_design()
Map: plot_augmented_row_column_design()
Analysis (fixed, random blocks, random entries, SpATS spatial): analyze_augmented_row_column_design()

Augmented p-rep designs

Multi-location trials in which a share of the entries is repeated at each location (Williams, Piepho and Whitaker, 2011), built by search or from alpha and alpha(2) arrays.

Plan: augmented_prep_parameters()
Design: augmented_prep_design(), augmented_prep_concurrence()
Map: plot_augmented_prep_design()
Analysis (with replicates and fixed or random genotype x location): analyze_augmented_prep_design()
Two-stage weighted analysis with block or SpATS spatial stage 1: analyze_augmented_prep_two_stage()

Resolvable row-column designs

Fully replicated designs optimized for row-column and spatial (linear variance) efficiency, with latinized options (Williams, John and Whitaker, 2006).

Design: resolvable_row_column_design()
Efficiency, neighbours, AR criterion: resolvable_row_column_efficiency()
Map: plot_resolvable_row_column_design()
Analysis (RC, LV + RC and AR + RC by REML): analyze_resolvable_row_column()

Saving to Breedbase

Designs are saved as trials with their plots into a breeding program and folder through BrAPI, after checking every name; after harvest, the phenotypic data are saved to the same plots.

Log in: breedbase_login()
Look up names: breedbase_programs(), breedbase_folders(), breedbase_locations()
Check location and accessions before designing: breedbase_check()
Convert designs generated with agricolae: as_breedbase_design()
Prepare an agricolae trial with its name, location and program: prepare_breedbase_trial()
Save (with a dry run first): save_breedbase()
Phenotypes: phenotype_template() (sheet with the plot names), save_breedbase_phenotypes() (values by plot and trait name)
Analysis results: first_stage_values() and save_breedbase_first_stage() (plot values as a trait of the field plots, trait made by breedbase_adjusted_trait()); second_stage_values() and save_breedbase_analysis() (genotype values as a Breedbase analysis)

Multi-environment trials from Breedbase

Read files: read_breedbase_file(), prepare_breedbase_pheno()
Error variances (Hartley's Fmax): run_breedbase_variance_check(), hartley_fmax_test()
Environment groups (Lin, 1982): run_environment_stratification(), lin_group_environments()
Maps: read_locations(), map_environment_groups()

Trait data and two-stage analysis

Look: plot_trait(); outliers: find_outliers(), remove_outliers()
Two-stage analysis for any design: analyze_two_stage(); single stage: analyze_single_stage()
Phenotypes of trials from Breedbase: breedbase_phenotypes()
Raw versus first-stage values: compare_first_stage()

Checks

check_setup(), check_design(), check_result(), check_summary(), prediction_accuracy().

Author

Maintainer: Christiano Simoes ccsimoes263@gmail.com

Authors:

Christiano Simoes ccsimoes263@gmail.com