How to replace NA values in dataframe with adjacent columns value + additional text to differentiate them in R

I'm curious if I could replace NA values in my data frame with text from the column to the left (that does not have NA), with an additional "_unclassified" text on the end.

Here is an example data frame:

feature <- c("1",
                  "2", 
                  "3", 
                  "4", 
                  "5" )
phylum <- c("Firmicutes",
                   "Firmicutes", 
                   "Firmicutes", 
                   "Proteobacteria", 
                   "Firmicutes" )
class <- c(NA,
                  "Clostridia", 
                  "Clostridia", 
                  "Gammaproteobacteria", 
                  "Bacilli" )
order <- c(NA,
                   NA,
                   "Oscillospirales", 
                   "Enterobacterales", 
                   "Staphylococcales" )
family <- c(NA,
                  NA,
                  NA,
                  "Enterobacteriaceae", 
                  "Staphylococcaceae" )
genus <- c(NA,
                  NA,
                  NA,
                  NA,
                  "Staphylococcus")


df <- data.frame(feature, phylum, class, order, family, genus)

For example,

feature 1 would have Firmicutes_unclassified across class, order, family, genus

feature 2 would have Clostridia_unclassified across order, family, and genus

feature 3 would have Oscillospirales_unclassified across family and genus

feature 4 would have Enterobacteriaceae_unclassified for genus

An option with na.locf from zoo

library(zoo)
df[-1] <- t(apply(df[-1], 1, \(x) ifelse(is.na(x), paste0(na.locf0(x), 
      '_unclassified'), x)))

-output

> df
  feature         phylum                   class                   order                       family
1       1     Firmicutes Firmicutes_unclassified Firmicutes_unclassified      Firmicutes_unclassified
2       2     Firmicutes              Clostridia Clostridia_unclassified      Clostridia_unclassified
3       3     Firmicutes              Clostridia         Oscillospirales Oscillospirales_unclassified
4       4 Proteobacteria     Gammaproteobacteria        Enterobacterales           Enterobacteriaceae
5       5     Firmicutes                 Bacilli        Staphylococcales            Staphylococcaceae
                            genus
1         Firmicutes_unclassified
2         Clostridia_unclassified
3    Oscillospirales_unclassified
4 Enterobacteriaceae_unclassified
5                  Staphylococcus

One-liner, using just base R.

df[-1] <- t(apply(df[-1], MARGIN=1, \(x) replace(x, is.na(x), paste0(tail(na.omit(x), n=1), '_unclassified'))))
df
#   feature         phylum                   class                   order                       family                           genus
# 1       1     Firmicutes Firmicutes_unclassified Firmicutes_unclassified      Firmicutes_unclassified         Firmicutes_unclassified
# 2       2     Firmicutes              Clostridia Clostridia_unclassified      Clostridia_unclassified         Clostridia_unclassified
# 3       3     Firmicutes              Clostridia         Oscillospirales Oscillospirales_unclassified    Oscillospirales_unclassified
# 4       4 Proteobacteria     Gammaproteobacteria        Enterobacterales           Enterobacteriaceae Enterobacteriaceae_unclassified
# 5       5     Firmicutes                 Bacilli        Staphylococcales            Staphylococcaceae                  Staphylococcus

Explanation:

We apply an anonymous function \(x) on MARGIN=1 (i.e. row-wise) on the data frame while excluding first column df[-1]. In the anonymous function we replace in every row x where is.na(x) is TRUE by the tail of length n=1 of na.omit(x) (i.e. x without the NAs while paste0ing suffix '_unclassified' to it.

Here is one potential solution which transposes the dataframe, fills the NAs with the most recent non-NA value, then transposes the dataframe back again:

library(tidyverse)
library(vctrs)

feature <- c("1", "2", "3", "4", "5" )
phylum <- c("Firmicutes", "Firmicutes", "Firmicutes", "Proteobacteria", "Firmicutes" )
class <- c(NA, "Clostridia", "Clostridia", "Gammaproteobacteria", "Bacilli" )
order <- c(NA, NA, "Oscillospirales", "Enterobacterales", "Staphylococcales" )
family <- c(NA, NA, NA, "Enterobacteriaceae", "Staphylococcaceae")
genus <- c(NA, NA, NA, NA, "Staphylococcus")


df <- data.frame(feature, phylum, class, order, family, genus)

df2 <- df %>%
  t() %>%
  as_tibble(.name_repair = ~vec_as_names(..., repair = "unique", quiet = TRUE)) %>%
  mutate(across(everything(), ~if_else(
    is.na(.x),
    paste0(vec_fill_missing(.x, direction = "down"), "_unclassified"),
    .x))) %>%
  t() %>%
  as_tibble(.name_repair = ~colnames(df))
df2
#> # A tibble: 5 × 6
#>   feature phylum         class                   order              family genus
#>   <chr>   <chr>          <chr>                   <chr>              <chr>  <chr>
#> 1 1       Firmicutes     Firmicutes_unclassified Firmicutes_unclas… Firmi… Firm…
#> 2 2       Firmicutes     Clostridia              Clostridia_unclas… Clost… Clos…
#> 3 3       Firmicutes     Clostridia              Oscillospirales    Oscil… Osci…
#> 4 4       Proteobacteria Gammaproteobacteria     Enterobacterales   Enter… Ente…
#> 5 5       Firmicutes     Bacilli                 Staphylococcales   Staph… Stap…

^{Created on 2023-02-10 with reprex v2.0.2}