Added functionality for writeBambuOutput() to write to gff3

[hafiz-ismail]

Related Issue

Fixes # (issue number)

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (please specify if the change breaks existing functionality)
  • Non breaking change (the feature doesn't change existing functionality)
  • Breaking change (the feature that would cause existing functionality to not work as expected)
• Documentation update
• Performance optimization

Description

Implemented functionality for writeBambuOutput to export gff3 files from SummarizedExperiment objects. Currently Bambu can export the more stringent gtf file format, but there is no functionality for gff3. An overview of the differences of the 2 file formats:

GTF differs from GFF3 mainly in the 9th column (attribute) and the syntax of the key:value pairs.

GTF:

gene_id "GENE1"; transcript_id "TX1"; gene_name "BRCA1";
gene_id "GENE1"; transcript_id "TX1"; exon_number "1";

GTF documentation (Ensembl)

GFF3:

ID=TX1;Parent=GENE1;Name=BRCA1
ID=exon1;Parent=TX1;exon_number=1

GFF3 documentation (Ensembl)

Also, GTF attributes have a trailing ; whereas GFF3 does not.

Other differences include

• GFF3 header
• GTF column 3 (feature) is constrained to lesser terms than GFF3

Implementation Details

All code changes were implemented in readWrite.R.

• Implemented two new helper functions: writeToGFF3 and writeAnnotationsToGFF3, following convention from already existing writeToGTF and writeAnnotationsToGTF
• Implemented new parameter outputAnnFormat in writeBambuOutput: takes either "gtf" (default) or "gff3" as arguments
• renamed transcript_gtffn to transcript_annfn
• Updated documentation in README and comment markers parsed by roxygen2

Impact of Changes

• All previous functionality is preserved, writeBambuOutput still exports to .gtf by default unless specified otherwise. It was found that the current writeBambuOutput in devel_pre_v4 is not fully functional - therefore a minimal function is provided to test results as below:

writeBambuOutput <- function(se, path, prefix = "", outputExtendedAnno = TRUE, 
                             outputAll = TRUE, outputBambuModels = TRUE, outputNovelOnly = TRUE, seperateSamples = FALSE,
                             outputAnnFormat = "gtf") {
  
  if (missing(se) | missing(path)) {
    stop("Both summarizedExperiment object from bambu and
            the path for the output files are required.")
  } else if (!outputAnnFormat %in% c("gtf", "gff3")) {
    stop("Please specify a valid annotation outputAnnFormat: 'gtf' or 'gff3', or leave blank to output to GTF by default.")
  } else {
    outdir <- paste0(path, "/")
    if (!dir.exists(outdir))
      dir.create(outdir, recursive = TRUE)
    
    transcript_grList <- rowRanges(se)
    prefix <- ifelse(prefix != "", paste0(prefix, "_"), "")
    transcript_annfn <- paste(outdir, prefix, sep = "")
    
    if (outputAnnFormat == "gtf") {
      gtf <- writeAnnotationsToGTF(annotation = transcript_grList,
                                   file = transcript_annfn, outputExtendedAnno = outputExtendedAnno, 
                                   outputAll = outputAll, outputBambuModels = outputBambuModels, outputNovelOnly = outputNovelOnly)
    }
    
    if (outputAnnFormat == "gff3") {
      gff3 <- writeAnnotationsToGFF3(annotation = transcript_grList,
                                     file = transcript_annfn, outputExtendedAnno = outputExtendedAnno, 
                                     outputAll = outputAll, outputBambuModels = outputBambuModels, outputNovelOnly = outputNovelOnly)
    }
  }
}

In addition, this output has not been updated for compatibility with importBambuResults - providing functionality for this will be associated with future implementations for the prepareAnnotations function

Checklist

• My code follows the style guidelines of this project.
• I have performed a self-review of my own code.
• I have commented my code, particularly in hard-to-understand areas.
• I have made corresponding changes to the documentation (vignettes, man pages).
• I have tested the code on a full dataset, and any differences have been described in the Impact of Changes section.