R script to manipulate data

eben.r

library(tidyverse)
library(data.table)
library(janitor)

csv <- fread("Ebenn_code_data_21Jul21_18.08.csv")

features <- names(csv)[-c(1,2,4)]

sample_names <- csv$Name %>% 
  gsub("_flye_[a-z\\_]*|_hybrid","",.) %>% 
  unique()

compare_list <- list()
i <- 0
for (sample in sample_names) {
  pilot_sample <- csv %>% 
    filter(grepl(sample, Name)) %>% 
    select(-2,-4) %>% 
    transpose() %>% 
    row_to_names(row_number = 1) %>% 
    mutate(features = features) %>% 
    select(features, contains("hybrid"), everything()) %>% 
    mutate_all(.funs = toupper) %>% 
    mutate(across(contains("flye"), .fns = function(x) x == cur_data()[2])) %>% 
    summarise(across(contains("flye"), sum)) %>% 
    mutate(sample = sample) %>%
    select(sample, everything()) %>% 
    setNames(c("sample","flye_accurate","flye_accurate_medaka","flye_accurate_medaka_pilon","flye_fast","flye_fast_medaka"))
  print(pilot_sample)
  i = i + 1
  compare_list[[i]] <- pilot_sample
}

compare_df <- bind_rows(compare_list) %>% 
  mutate(across(contains("flye"), .fns = function(x) x/99))