Wilcoxon Rank Sum for Shikha

example_wilcox.R

# load libraries individually
library(purrr)
library(magrittr)
library(tidyr)
library(dplyr)

# or just load
library(tidyverse)

# some random data to make my own data frame with 
# the same structure as yours
set.seed(0)
df <- tibble(Protein.Id = letters[1:5],
             Gene.Symbol = LETTERS[1:5],
             Description = letters[1:5],
             Number.of.peptides = sample(1:20, 5),
             C1_G12D_vitro_1 = rnorm(5, 4, 2),
             C1_G12D_vitro_2 = rnorm(5, 4, 2),
             C1_G12D_vitro_3 = rnorm(5, 4, 2),
             C1_WT_vitro_1 = rnorm(5, 6, 2),
             C1_WT_vitro_2 = rnorm(5, 6, 2),
             C1_WT_vitro_3 = rnorm(5, 6, 2),
             C1_G12D_vivo_1 = rnorm(5, 4, 2),
             C1_G12D_vivo_2 = rnorm(5, 4, 2),
             C1_G12D_vivo_3 = rnorm(5, 4, 2),
             C1_WT_vivo_1 = rnorm(5, 6, 2),
             C1_WT_vivo_2 = rnorm(5, 6, 2),
             C1_WT_vivo_3 = rnorm(5, 6, 2))

# a wrapper around 'wilcox.test()' to take one row at a time
wrap_wilcox_test <- function(Gene.Symbol, data) {
    g12d <- unlist(data[, 1:3])  # G12D sample data
    wt <- unlist(data[, 4:6])    # WT sample data
    # run test and return p-value
    return(wilcox.test(g12d, wt)$p.value)
}

# get p-values for "in vitro" data
df$vitro_pval <- df %>%
    dplyr::select(Gene.Symbol, contains("vitro")) %>%
    dplyr::group_by(Gene.Symbol) %>%
    tidyr::nest() %>%
    purrr::pmap(wrap_wilcox_test)

# get p-values for "in vivo" data
df$vivo_pval <- df %>%
    dplyr::select(Gene.Symbol, contains("vivo")) %>%
    dplyr::group_by(Gene.Symbol) %>%
    tidyr::nest() %>%
    purrr::pmap(wrap_wilcox_test)
    
# There should now be 2 new columns "vitro_pval" and "vivo_pval"
#  in your data frame

# The two steps above use the following workflow:
#   1. select the gene symbol and either vitro or vivo columns
#   2. group the data by gene (i.e. each row is separate)
#   3. make each group (i.e. row, i.e. gene) into a separate data frame
#   4. run each through the wilcox.test wrapper function

# From here, adjusting for multiple hypothesis testing (with BH method) is easy
df <- df %>%
  mutate(vitro_pval_adj = p.adjust(vitro_pval, method = "BH"),
         vivo_pval_adj = p.adjust(vivo_pval, method = "BH"))

# This should create the columns "vivo_pval_adj" and "vivo_pval_adj"