R/get_original_peptide.R

In leonjessen/Barracoda2viz:

get_original_peptide = function(x){
  if(!is.tibble(x)){
    stop("Input to clean_barracoda_data is not a tibble")
  }
  aa = c('A','R','N','D','C','Q','E','G','H','I',
         'L','K','M','F','P','S','T','W','Y','V')
  originals = tibble()
  samples = x %>% pull(sample) %>% unique
  for( id in samples ){
    tmp = x %>% filter(sample==id,!str_detect(pep.number,"original")) %>% get_mut_pos %>% get_mut_res %>%
      select(mut_pos,mut_res_one)
    positions = tmp %>% pull(mut_pos) %>% unique %>% sort
    org = ''
    for( p_i in positions ){
      substitions = tmp %>% filter(mut_pos == p_i) %>% pull(mut_res_one)
      org_res = aa[!aa %in% substitions]
      if( length(org_res) > 1 ){
        warning(paste0("At peptide position '",p_i,"' in sample '",id,
                       "' the residue could not be unambiguously determined. ",
                       "Possibilities are '",paste(org_res, collapse = ','),
                       "'. Note, 'X' is assigned!"))
        org_res = 'X'
      }
      org = paste0(org,org_res)
    }
    originals = originals %>% bind_rows(tibble(sample=id,original=org))
  }
  return(originals)
}