Chaining target factories: target not found

[prototaxites]

Help

• I understand and agree to https://books.ropensci.org/targets/help.html.

Description

Trying to build a set of target factories to abstract away some repetitive tasks I'll need to run a lot. I am quite new to the concept of metaprogramming, so there's probably something obvious I've missed.

There are two factories: the first (tar_read_ampliseq) reads in the data and creates an output (phyloseq) object, and the second (tar_decontaminate_phyloseq) does some processing of this object downstream. However, the second factory gives me an error saying the object is not found. Running the first target factory on its own works, and the target is listed in tar_manifest(). I can also access the target subsequently with a call to tar_target() - just not my second factory:

Target factories:

tar_read_ampliseq <- function(name, table, taxonomy, metadata) {
  tar_name <- deparse(substitute(name))
  name_table_file <- paste0(tar_name, "_table_file")
  name_taxonomy_file <- paste0(tar_name, "_taxonomy_file")
  name_metadata_file <- paste0(tar_name, "_metadata_file")
  name_table <- paste0(tar_name, "_table")
  name_taxonomy <- paste0(tar_name, "_taxonomy")
  name_metadata <- paste0(tar_name, "_metadata")
  name_phyloseq <- paste0(tar_name, "_phyloseq")
  list(
    tar_target_raw(name_table_file, table, format = "file"),
    tar_target_raw(name_taxonomy_file, taxonomy, format = "file"),
    tar_target_raw(name_metadata_file, metadata, format = "file"),
    tar_target_raw(name_table, 
                   substitute(read_tsv(file, col_types = cols()),
                   env = list(file = as.symbol(name_table_file)))),
    tar_target_raw(name_taxonomy, 
                   substitute(read_tsv(file, col_types = cols()),
                              env = list(file = as.symbol(name_taxonomy_file)))),
    tar_target_raw(name_metadata, 
                   substitute(read_tsv(file, col_types = cols()),
                              env = list(file = as.symbol(name_metadata_file)))),
    tar_target_raw(name_phyloseq, substitute(
      build_phyloseq(table, taxonomy, metadata),
      env = list(table = as.symbol(name_table),
                 taxonomy = as.symbol(name_taxonomy),
                 metadata = as.symbol(name_metadata))
      )
      )
  )
}

tar_decontaminate_phyloseq <- function(phyloseq, neg = NULL, batch = NULL) {
  if(is.null(neg)) { stop("No column defining negative controls was provided!") }
  name_contaminants <- paste0(deparse(quote(phyloseq)), "_contaminants")
  name_decontaminated <- paste0(deparse(quote(phyloseq)), "_decontaminated")
  
  list(
    tar_target_raw(name_contaminants,
                   substitute(
                     decontam::isContaminant(phylo, neg = neg, batch = batch, method = "frequency"),
                     list(phylo = phyloseq, neg = neg, batch = batch)
                   )),
    tar_target_raw(name_decontaminated,
                   substitute(
                     prune_taxa(!contaminants$contaminant, phylo),
                     list(phylo = phyloseq, contaminants = as.symbol(name_contaminants))
                   ))
  )
}

_targets.R file:

library(targets)
tar_option_set(packages = c("tidyverse", "phyloseq"))

list(
  ## Read in data
  tar_read_ampliseq(rrna_16s, 
               here::here("path/to/file"),
               here::here("path/to/file"),
               here::here("path/to/file")
               ),
  tar_decontaminate_phyloseq(rrna_16s_phyloseq, "Control")
)

Error:
! Error running targets::tar_make()
  Error messages: targets::tar_meta(fields = error, complete_only = TRUE)
  Debugging guide: https://books.ropensci.org/targets/debugging.html
  How to ask for help: https://books.ropensci.org/targets/help.html
  Last error: object 'rrna_16s_phyloseq' not found

If anyone has any insight, it would be much appreciated - I can't see for the life of me what I've missed!

[wlandau]

Looks like one issue is how tar_decontaminate_phyloseq() captures the target name. I can reproduce it outside the pipeline:

tar_decontaminate_phyloseq(rrna_16s_phyloseq, "Control")
#> Error: object 'rrna_16s_phyloseq' not found

If you want to pass the name as a symbol, deparse(substitute()) will quote the argument and turn it into a string.

tar_decontaminate_phyloseq <- function(phyloseq, neg = NULL, batch = NULL) {
  target_name <- deparse(substitute(phyloseq))
  ...
}

Once you fix this, I recommend checking tar_visnetwork() to see if the dependency graph is correct.

Unfortunately I cannot test the pipeline because I do not have the phyloseq package, and I do not have the physical files it seems to require. If you have more trouble, it would help if the example code were simpler and runnable, as in a reprex.

[prototaxites]

Apologies for the late reply to this, I had to put this down for a couple of weeks. Thanks for the tips, and apologies for the non-reproducable code - I'll make sure to build a better example next time :)

I managed to fix this today - I switched out the code as you suggested above, which didn't fix things. What I had to do was change substitute(..., list(phylo = phyloseq)) to substitute(..., list(phylo = as.symbol(phyloseq))), and now the function works.

I definitely need to brush up on my R metaprogramming skills before I can progress much further with this - can you recommend any good resources for learning the basics?

[wlandau]

For meta programming, I have some pointers at https://wlandau.github.io/targetopia/contributing.html. https://adv-r.hadley.nz/expressions.html is more comprehensive.