I've updated the function to convert the "mapped" column output to:

1. Convert from a list to character. This makes the dataframe writeable to csv.

2. Set a ; delimiter for the GO terms column to simplify downstream parsing.

3. Update the column name (just a personal preference).

mappedIds <-
  function(df, collection, OFFSPRING)
  {
    map <- as.list(OFFSPRING[rownames(df)])
    mapped <- lapply(map, intersect, ids(collection))
    df[["go_terms"]] <- vapply(unname(mapped), paste, collapse = ";", character(1L))
    df
  }

Those don't look like standard gene ids, but maybe trinity de novo transcript identifiers? That probably deserves a new question -- 'how to map from ...?'

Probably, the solution involves the GeneSetCollection class and constructor. This code is adapted from the example of its help page:

lst <- list(GOterm1=c("TRINITY_g1", "TRINITY_g2"),
            GOterm2=c("TRINITY_g2", "TRINITY_g5"))
gsc <- GeneSetCollection(mapply(function(geneIds, GOID) {
         GeneSet(geneIds, collectionType=GOCollection(GOID), setName=GOID)
       }, lst, names(lst)))
gsc
GeneSetCollection
  names: GOterm1, GOterm2 (2 total)
  unique identifiers: TRINITY_g1, TRINITY_g2, TRINITY_g5 (3 total)
  types in collection:
    geneIdType: NullIdentifier (1 total)
    collectionType: GOCollection (1 total)

If you can use standard gene identifiers such as Entrez, then you would add the argument shown in the help page geneIdType=EntrezIdentifier() to the call to GeneSet. If these are novel transcripts assembled with trinity, as your sample input suggests, then you are not in that situation.

Alrighty, figured this out! Here's the full code, including the initial mapping step described in the accepted answer above, and subsequent steps to get GO slims to map to themselves.

# Vector of GO IDs
go_ids <- pgen_GO_gene_mappings_table_cleaned_separated_grouped_df$all_GO_ids

# Create GSEAbase GOCollection using `go_ids`
myCollection <- GOCollection(go_ids)

# Retrieve GOslims from GO OBO file set
slim <- getOBOCollection(gseabase_files)

# Retrieve Biological Process (BP) GOslims
slimdf <- goSlim(myCollection, slim, "BP", verbose)

# List of GOslims and all GO IDs from `go_ids`
gomap <- as.list(GOBPOFFSPRING[rownames(slimdf)])

# Maps `go_ids` to matching GOslims
mapped <- lapply(gomap, intersect, ids(myCollection))

# Append all mapped GO IDs to `slimdf`
# `sapply` needed to apply paste() to create semi-colon delimited values
slimdf$ids <- sapply(lapply(gomap, intersect, ids(myCollection)), paste, collapse=";")

# Remove "character(0) string from "ids" column
slimdf$ids[slimdf$ids == "character(0)"] <- ""

# Add self-matching GOIDs to "ids" column, if not present
for (go_id in go_ids) {
  # Check if the go_id is present in the row names
  if (go_id %in% rownames(updated_slimdf)) {
    # Check if the go_id is not present in the ids column
    # Also removes white space "trimws()" and converts all to upper case to handle
    # any weird, "invisible" formatting issues.
    if (!go_id %in% trimws(toupper(strsplit(updated_slimdf[go_id, "ids"], ";")[[1]]))) {
      # Append the go_id to the ids column with a semi-colon separator
      if (length(updated_slimdf$ids) > 0 && nchar(updated_slimdf$ids[nrow(updated_slimdf)]) > 0) {
        updated_slimdf[go_id, "ids"] <- paste0(updated_slimdf[go_id, "ids"], "; ", go_id)
      } else {
        updated_slimdf[go_id, "ids"] <- go_id
      }
    }
  }
}

The resulting output file looks like this: