natfeedback_code_eLife/R/get_data.R

get_data <- function(csv_file_name) {
  # Reads data from csv file into a tibble, and creates IDs for mice, series, experiments, and neurons
  
  tib = read_csv(csv_file_name)
  
  if (has_name(tib, 'mseu')) {
  
    # add mouse ID - this is based on m(ouse) only
    tib <- tib %>% mutate(mid = as.numeric(as.factor(str_split_fixed(tib$mseu, '_s', 2)[,1])))
    
    # add series ID - this is based on m and s(eries)
    tib <- tib %>% mutate(sid = as.numeric(as.factor(str_split_fixed(tib$mseu, '_e', 2)[,1])))
    
    # add exp ID - this is based on m, s, and e(xperiment)
    tib <- tib %>% mutate(eid = as.numeric(as.factor(str_split_fixed(tib$mseu, '_u', 2)[,1])))
    
    # add unit ID - this is based on m, s, and u(nit), ignoring e
    tib <- tib %>% mutate(uid = as.numeric(as.factor(paste(str_split_fixed(tib$mseu, '_e', 2)[,1], '_u' , str_split_fixed(tib$mseu, '_u', 2)[,2], sep=""))))
    
    # put iDs up front
    tib <- tib %>% select(mid, sid, eid, uid, everything())
  
    # If we don't have 'experiments', we need to do a slightly different parsing
  } else if (has_name(tib, 'msu')) {
    
    # add mouse ID - this is based on m only
    tib <- tib %>% mutate(mid = as.numeric(as.factor(str_split_fixed(tib$msu, '_s', 2)[,1])))
    
    # add series ID - this is based on m and s
    tib <- tib %>% mutate(sid = as.numeric(as.factor(str_split_fixed(tib$msu, '_u', 2)[,1])))
    
    # add unit ID - this is based on m, s, and u
    tib <- tib %>% mutate(uid = as.numeric(as.factor(tib$msu)))
    
    # put IDs up front
    tib <- tib %>% select(mid, sid, uid, everything()) # put id up front
    
    # We don't have units - this requires a slightly differnt parsing
  } else if (has_name(tib, 'mse')) {
    
    # add mouse ID
    tib <- tib %>% mutate(mid = as.numeric(as.factor(str_split_fixed(tib$mse, '_s', 2)[,1])))
    
    # add series ID
    tib <- tib %>% mutate(sid = as.numeric(as.factor(str_split_fixed(tib$mse, '_e', 2)[,1])))
    
    # add experiment ID
    tib <- tib %>% mutate(eid = as.numeric(as.factor(tib$mse)))
    
    # put IDs up front
    tib <- tib %>% select(mid, sid, eid, everything()) 
    
  }

  return(tib)
}