read_gff3 can't parse when exons are defined and CDS is called 'cds' instead of 'CDS'

[henni164]

There seems to be some kind of parsing bug (well, maybe not a bug per se) that causes read_gff3() to fail when trying to map exon and CDS information if the feature type label for CDS is lowercase instead of uppercase. If only gene/mRNA/cds features are present, it does not matter if the cds feature label is lowercase or uppercase.

Running R4.4.2 in RStudio for Windows

Full error message:

gff <- read_feats("C:/Users/hen294/my.gff", format = "gff3")
Reading 'gff3' with read_gff3():

• file_id: my [C:/Users/hen294/my.gff]
  Harmonizing attribute names
  • ID -> feat_id
  • Parent -> parent_ids
  Error in map2():
  ℹ In index: 1.
  ℹ With name: my.
  Caused by error in bind_rows():
  ! Can't combine ..1$feat_id and ..2$feat_id .
  Run rlang::last_trace() to see where the error occurred.

rlang::last_trace()
<error/purrr_error_indexed>
Error in map2():
ℹ In index: 1.
ℹ With name: my.
Caused by error in bind_rows():
! Can't combine ..1$feat_id and ..2$feat_id .

---

Backtrace:
▆

1. ├─gggenomes::read_feats(...)
2. │ └─gggenomes::read_context(...)
3. │ └─purrr::map2_df(...)
4. │ └─purrr::map2(.x, .y, .f, ...)
5. │ └─purrr:::map2_("list", .x, .y, .f, ..., .progress = .progress)
6. │ ├─purrr:::with_indexed_errors(...)
7. │ │ └─base::withCallingHandlers(...)
8. │ ├─purrr:::call_with_cleanup(...)
9. │ └─gggenomes (local) .f(.x[[i]], .y[[i]], ...)
10. │ ├─rlang::exec(parser, file, ...)
11. │ └─gggenomes::read_gff3("C:/Users/hen294/my.gff")
12. │ ├─bind_rows(mrna_exon_introns, mrna_cds_introns) %>% ...
13. │ └─dplyr::bind_rows(mrna_exon_introns, mrna_cds_introns)
14. └─dplyr::mutate(., feat_id = as.character(.data$feat_id))

Minor issue but may be worth including case insensitivity for the feature type column.

[thackl]

Thanks, good point. Will add that in the next version.

For gff2/gtf file, I even do it already. If you need a quick fix for your local installation of gggenomes, just edit read_gff3.R and change the line 66 to if(TRUE){ to always apply the case conversion.

gggenomes/R/read_gff3.R

Lines 66 to 81 in 0cdd5e7

	if (is_gff2) {
	warn(str_glue(
	"This looks like a gff2/gtf file. This is usually fine, ",
	"but given the ambigious definition of this format, it is not ",
	"guaranteed that gene models are always captured correctly. ",
	"exons/CDS might not be recognized as belonging to the same gene, etc. ",
	"Also note: types and attributes are as far as possible converted to ",
	"match gff3 standards (transcript -> mRNA, 5'/3'UTR -> five/three_prime_UTR, ...)"
	))
	# make this consistent with gff3
	x[[3]][x[[3]] == "transcript"] <- "mRNA"
	x[[3]][x[[3]] == "5'UTR"] <- "five_prime_UTR"
	x[[3]][x[[3]] == "3'UTR"] <- "three_prime_UTR"
	x[[3]][x[[3]] == "cds"] <- "CDS"
	}

[henni164]

Thank you! My fix was just to sed replace cds with CDS in my gff, but it did take me quite a while to figure out why adding exons to my file was breaking read_gff3(), so it will be good to have that updated. :)