schmidtpaul/dsfair

archived: Data Science for Agriculture in R; has been succeeded by https://schmidtpaul.github.io/dsfair_quarto/

DSFAIR - Data Science for Agriculture in R

📊🌱 DSFAIR - Data Science for Agriculture in R provides a cookbook with statistical analyses of typical examples in life sciences (focus on agriculture/biology). It can be seen as the smaller brother of 📈🌱 MMFAIR - Mixed Models for Agriculture in R, because it deals with more basic analyses.

Where possible, I try to use example data/analyses that have been published in university lecture notes or peer-reviewed journals.

Note that 📉🌱 R für Bio- und Agrarwissenschaftler is the German predecessor of this website.

dsfair's Issues

Github Site Content To Do List

It would be nice to have these things in the future:

Chapters

Add complete chapters on this

• simple linear regression 📑 ❓
• correlation 📑 ❓
• outliers 📑 ❓
• how to prepare for online courses
  • install R & RStudio
  • update R & RStudio
  • download & unzip data/resource folder
• how to reprex/stackoverflow 📑 ❓
• how to export 📑 ❓
  • broom/broom.mixed
  • openxlsx quick/proper
  • format p-values
  • ggsave/gg_export
• polynomial regression 📑 ❓
• actually helpful covariates 📑 ❓
• better example for subsampling - maybe pigs?
• discussion on p-values and significance (german version)
• discussion on t-test, Tukey-test (german version)
• discussion on interactions (german version)
• discussion on model diagnostics/residuals, transformations and generalized models (german version)
• discussion on random effects, mixed models and covariance structures (german version, also this and this)
• discussion on adjusted means cite this
• discussion on contrasts between (adjusted) means cite this
• discussion on compact letter display
• discussion on Coefficient of Variance (CV)
• Rmarkdown
  • useful links for the first time setup for PDF and Word
  • useful links for using it
• Github
  • useful links to packages etc.
  • useful links to getting started with git in R

Examples

Add DIY examples at the end of every chapter

• check which ones are missing
• check if the purl-generated solutions are fine
• better example for 1f RCBD (not density as pseudo factor)

Other

• consistency for loaded packages across chapters
• most often used ggplot tips
• ggplot captions using ggtext
• replace _at & _if with across()

emmeans S4 object issue

Copy pasted this from an email:

Thanks for developing the test examples using DSFAIR! I ran the code you made available on https://schmidtpaul.github.io/DSFAIR/augmented_Pattersen1994.html However, i found several challenges in getting it to work with recent R version (3.6.3 and above) when it comes to use the “emmeans” function, specifically this section

mean_comparisons <- mod.fb %>%
  emmeans(pairwise ~ "gen",
          adjust = "tukey") %>%
  pluck("emmeans") %>%
  cld(details = TRUE, Letters = letters) # add letter display

# If cld() does not work, try CLD() instead.
# Add 'adjust="none"' to the emmeans() and cld() statement
# in order to obtain t-test instead of Tukey!

mean_comparisons$emmeans # adjusted genotype means

For unclear reason the provided code work just fine with some older R distributions (3.6) but with most recent ones it would throw a S4 class error (it is an emmGrid object) when running: mean_comparisons$emmeans, such as:

#> mean_comparisons$emmeans # adjusted genotype means
#> Error in mean_comparisons$emmeans : 
#>   $ operator not defined for this S4 class

I could get around problem converting the object using getSlot function but which return different data structure from “mean_comparisons”. Wondered if you are aware of such issue and if you could suggest a more elegant way to get around? I am running R on a mac, using Rswitch which includes R version 3.6, 3.6.3, 4.0.2

If I decide to drop those markers that are present in less than 2% of lines, or equivalently those that are present in 98% of lines, because they behave almost as constants. How do we do that?

Hello sir
I am facing an issue in multivariant to remove the 2% of the markers from lines...
here is my initial code.

LDA

clean start

rm(list=ls())

installing and loading specialized libraries

library(lattice)
tryCatch({library(ggplot2)}, error = function(e) {install.packages("ggplot2")},finaly = library(ggplot2))
tryCatch({library(MASS)}, error = function(e) {install.packages("MASS")},finaly = library(MASS))
tryCatch({library(caret)}, error = function(e) {install.packages("caret")},finaly = library(caret))
#tryCatch({library(tidyverse)}, error = function(e) {install.packages("tidyverse")},finaly = library(tidyverse))

#library(gmodels)

Are you in the appropriate working directory?

getwd()

were are the test data

dir("Data/")

#reading data
Data=read.table("SunFlowerGenotypes.txt",header = T,sep="\t")

#showing header of data
head(Data)
colnames(Data)

#List of proportions line have
sapply(Data[,-1],mean)

#Eleminate the 2% of lines from markers
sapply(Data[,-1],mean)<0.02

I need help in to remove the 2% of markers from lines.
Here is the data
SunFlowerGenotypes.txt
I will be happy if you help me to figure out the code