LMSstat
Package for automation of statistics that are widely used in metabolomics.
Install / Use
/learn @CHKim5/LMSstatREADME
LMSstat
Automation of statistical test with an identical data input aiming to reduce arduous work searching for packages and changing data input.
The package includes
-
Simple Statistics :u-test, t-test, post hocs of Anova and Kruskal Wallis with FDR adjusted values
-
Bar, Box, Dot, Violin plots with significance (u-test, t-test, post hocs of Anova and Kruskal Wallis)
-
Scaling & Transformation
-
Normality check (Shapiro Wilk test)
-
Scheirer–Ray–Hare Test
-
Volcano plot
-
Heatmap
-
PERMANOVA
-
NMDS
-
PCA
-
PCoA
Contribution acknowledgement
Oct.01/2021 Daehwan Kim
-
Allstats_new optimization for faster processing
-
bug fix of Allstats (regarding LETTERS210729)
Instructions
Installation
Download R
https://cran.r-project.org/bin/windows/base/
Download R Studio
https://www.rstudio.com/products/rstudio/download/
Download Rtools
https://cran.r-project.org/bin/windows/Rtools/
Download package in R
install.packages("devtools")
devtools::install_github("CHKim5/LMSstat")
library(LMSstat)
Basic structure of the Data
Used in
- Simple statistics
- Barplot, Boxplot, Dotplot
- Volcano plot
- Scheirer–Ray–Hare Test
- PERMANOVA
- NMDS
- PCA
- Scaling & Transformation
- Normality check (Shapiro Wilk test)
- Heatmap
#Sample Data provided within the package
data("Data")
# Uploading your own Data
setwd("C:/Users/82102/Desktop")
Data<-read.csv("statT.csv",header = F)
The column "Multilevel" is mandatory for the code to run flawlessly.
If Multilevel is not used, fill the column with random characters
Datafile needs to follow the following format
Care for Capitals: Sample, Multilevel, Group
<p align="center"> <img src="https://user-images.githubusercontent.com/77651662/125877205-e140d306-81d8-459f-8414-b8ef3bca63d7.PNG" width="750" height="400"> </p> <p align="center">statT.csv</p>Used in
- PERMANOVA
#Sample Data provided within the package
data("Classification")
# Uploading your own Data
Classification<-read.csv("statT_G.csv",header = F)
Univariate statistics
Statfile<-Allstats_new(Data,Adjust_p_value = T, Adjust_method = "BH") # Optimized code using lapply / data.table for faster processing contributed by Daehwan Kim
Statfile<-Allstats(Data,Adjust_p_value = T, Adjust_method = "BH") # Previous version using for-loop
Adjustable parameters
-
Adjust_p_value = T # Set True if adjustment is needed
-
Adjust_method = F # Adjustment methods frequently used. c("holm", "hochberg", "hommel", "bonferroni", "BH", "BY","fdr", "none")
head(Statfile[["Result"]]) # includes all statistical results
write.csv(Statfile[["Result"]],"p_value_result.csv") # Write csv with all the p-value included
Plots
# Makes a subdirectory and saves box plots for all the variables
AS_boxplot(Statfile,asterisk = "u_test")
# Makes a subdirectory and saves dot plots for all the variables
AS_dotplot(Statfile,asterisk = "t_test")
# Makes a subdirectory and saves bar plots for all the variables
AS_barplot(Statfile,asterisk = "Scheffe")
# Makes a subdirectory and saves violin plots for all the variables
AS_violinplot(Statfile,asterisk = "Scheffe")
AS_boxplot(Statfile) AS_dotplot(Statfile)
<p align="center"> <img src="https://user-images.githubusercontent.com/77651662/125714687-1908d6eb-b2bd-4e25-8ef0-62c24466c32a.png" width="350" height="250"> <img src="https://user-images.githubusercontent.com/77651662/125714704-d7dab67e-03c0-4e35-b86a-36723f7c63de.png" width="350" height="250"> </p>AS_barplot(Statfile) AS_violinplot(Statfile)
<p align="center"> <img src="https://user-images.githubusercontent.com/77651662/125715925-0878ec77-30bf-4859-8e56-316d98b6d520.jpg" width="350" height="250"> <img src="https://user-images.githubusercontent.com/77651662/141611647-1bc0f049-fcd1-4cb4-a713-5aa30f24dcb1.png" width="350" height="250"> </p>Adjustable parameters
- asterisk = "t_test" #c("Dunn","Scheffe","u_test","t_test")
- significant_variable_only = F # If set to TRUE, insignificant results will not be plotted
- color = c("#FF3300", "#FF6600", "#FFCC00", "#99CC00", "#0066CC", "#660099") # Colors for the plots
- legend_position = "none" # "none","left","right","bottom","top"
- order = NULL # Order of the groups c("LAC","LUE","WEI","SDF","HGH","ASH")
- tip_length = 0.01 # significance tip length
- label_size = 2.88 # significance label size
- step_increase = 0.05 #significance step increase
- width = 0.3 # box width ; size = 3 # dot size
- fig_width = NA #figure size
- fig_height = NA #figure size
- Y_text = 12 # Y title size
- X_text = 10 # X text size
- Y_lab = 10 #y axis text size
- T_size = 15 # Title size
- sig_int = c(0.1,0.05) # significance interval
Scaling & Transformation
scaled_data<-D_tran(Data,param = "Auto")
Raw_Data Scaled_Data
<p align="center"> <img src="https://user-images.githubusercontent.com/77651662/126724661-1e2ee121-ecfe-41eb-a52a-e3dca128c12c.PNG" width="400" height="200"> <img src="https://user-images.githubusercontent.com/77651662/126724676-44c4eaac-c007-4eaf-9962-0b739291adb6.PNG" width="400" height="200"> </p>Adjustable parameters
-
param = "None" # "None","Auto","log10","Pareto"
-
save = F #Set true if datafile is to be saved
Normality check
#Shapiro Wilk test
Result<-Norm_test(Data)
write.csv(Result,"Normality_test_Result.csv")
Scheirer–Ray–Hare Test
# csv files including significant variables (Multilevel, Group, interaction) and a Venn diagram are downloaded
SRH(Data)
Adjustable parameters
- Adjust_p_value = T # Set True if adjustment is needed
- Adjust_method = "BH" # Adjustment methods frequently used. c("holm", "hochberg", "hommel", "bonferroni", "BH", "BY","fdr", "none")
Volcano plot
# Makes a subdirectory and saves Volcano plots for different combination of groups
Test<-Allstats(Data)
Volcano(Test,asterisk = "t-test")
Adjustable parameters
- asterisk = "t-test" #statistics inheriting from Allstats "Scheffe", "t-test", "u-test", "Dunn"
- reverse = T # T, F reverse the direction of fold change
- fig_width = NA #figure size
- fig_height = NA #figure size
- FC_log = 2 # Fold change log transformation value
- pval_log = 10 #p_value log transformation value
- dotsize = 3 #dotsize
- x_limit = c(-2,2) #x axis limt
- y_limit =c(0,6) #y axis limit
- pval_intercept = 0.05 # intercept for identification
- sig_label = T # T,F label significant variables
- color=c("#FF3300","#FF6600","#FFCC00") #colors used for ggplots.
- fixed_limit = F #whether the limit should be fixed or not T, F
- max_overlap = 20 #maximum overlap for labels
- FC_range = c(-1.5,1.5) #significant fold change range
Heatmap
# Makes a subdirectory and saves Heatmap
scaled_data<-D_tran(Data,param = "Auto")
AS_heatmap(scaled_data) #data inheriting from D_tran
dev.off() # Saved as PDF
Adjustable parameters
- col =c("green", "white", "red") # colors for heatmap
- col_lim = c(-3, 0, 3) # color boundaries
- reverse = T # T,F Reverse column and rows
- distance = "pearson" # Distance matrix for HCA "pearson", "manhattan","euclidean","spearman","kendall" ,
- rownames = T # T,F
- colnames = T # T,F
- Hsize = (3,6) # Width & Height c(a,b)
- g_legend = "Group" # Annotation legend title
- h_legend = "Color Key" # Heatmap legend title
- Title ="Title" # Title
- T_size = 10 # Title text size
- R_size = 3 # row text size
- C_size = 3 # column text size
- Gcol =c("ASD" = "black","HGH"="red","LAC"="blue","LUE" ="grey","SDF" = "yellow","WEI"="green") # Color for top_annotation bar
- dend_h = 0.5 #dendrite height
- a_h = 0.2 # top annotation hegiht
Multivariate statistics
PERMANOVA
data("Data")
data("Classification")
Single factor
PERMANOVA done with the Group column
Indiv_Perm(Data) # The group information is treated as a factor
Multiple Factors
Loops PERMANOVA over different classes provided by Classification
Result<-Multi_Perm(Data,Classification) # The group information is treated as factors
Adjustable parameters
- method = Dissimilarity index c("manhattan", "e
