.Rhistory

rm(list = ls())
library(openxlsx)
library(tidyverse)
library(car)
library(pheatmap)
library(broom)
library(ggpubr)
library(viridisLite)
library(modelr)
#library(dlookr)
#library(imputeLCMD)
library(ggrepel)
here::i_am("workflows/GC_MS_normalization/210927_primary_normalization_with_split.R")
library(here)
out <- here("runs/GC-MS normalization")
if (file.exists(out)) {
cat("The folder already exists")
} else {
dir.create(out)
}
setwd(here())# Not recommended but convenient in Rstudio to start from root
sam_dat1 <- readxl::read_xlsx(here("studies/cmQTL_val1_GH_2020/isa.study.xlsx"))
View(sam_dat1)
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = str_extract(`Factor [sample fresh weight]`, "^\\d{2},\\d{2}"))
View(sam_dat1_tidy)
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = str_extract(`Factor [sample fresh weight]`, "\\d{2},\\d{2}"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}")))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(str_remove(Parameter))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(str_remove("Parameter"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(str_remove(pattern = "Parameter"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(str_remove(string = .x, pattern = "Parameter"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), str_remove(string = .x, pattern = "Parameter"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = str_remove(string = .x, pattern = "Parameter"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter"))
View(sam_dat1_tidy)
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "\\w"), replacement = "_")
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "\\w", replacement = "_"))
?stringr
View(sam_dat1_tidy)
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "\\s", replacement = "_"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "[:blank:]", replacement = "_"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter [")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic [")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "[:blank:]", replacement = "_"))
sam_dat1_tidy <- GC_run1 %>%
left_join(GC_machine_nums) %>%
select(extraction_num = `Sample name`, everything())%>%
mutate(class = as_factor(if_else(str_detect(extraction_num, "run_qc"), "run_qc",
if_else(str_detect(extraction_num, "blank"), "blank", "sample"))),
extraction_num = as.numeric(if_else(str_detect(extraction_num, "run_qc"), "0",
if_else(str_detect(extraction_num, "blank"), "-1",extraction_num))),
exp = as_factor(1)) %>%
left_join(sam_dat1) %>%
left_join(genotypes) %>%
select(treatment = irrigation, everything()) %>%
select(all_of(sam_vars)) %>%
arrange(run_num_GC)
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter [")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic [")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "[:blank:]", replacement = "_"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter \\[")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic \\[")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "[:blank:]", replacement = "_"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter \\[")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic \\[")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Factor \\[")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "[:blank:]", replacement = "_"))
sam_dat1_tidy <- sam_dat1 %>%
mutate(sample_weight = as.double(str_extract(`Factor [sample fresh weight]`, "\\d{2}\\.\\d{2}"))) %>%
rename(source_name = `Source Name`) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Parameter \\[")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Characteristic \\[")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "Factor \\[")) %>%
rename_with(.cols = everything(), .fn = ~str_remove(string = .x, pattern = "\\]")) %>%
rename_with(.cols = everything(), .fn = ~str_replace_all(string = .x, pattern = "[:blank:]", replacement = "_"))
x <- c(1,2,3,4)
y <- c("A", "B","C", "D")
replace(x, c(3,2,1,4), y)
colnames(sam_dat1_tidy)
sam_dat1_tidy$`multi-tissue_plant_structure`