Improvement proposal for gtf2featureAnnotation.R (GTF2GeneList tool)

[cgirardot]

Hi,
I know the tool is for ensemble GTF and I am using it on mouse gencode GTF... It fails on line 141-143 of gtf2featureAnnotation.R when asking to flag Mitochondrial features.

if (opt$mito){
  anno$mito <- ucfirst(as.character(tolower(anno$gene_biotype) %in% cleanlist(opt$mito_biotypes) | tolower(anno$chromosome) %in% cleanlist(opt$mito_chr)))
}

this is because the script assumes gene_biotype is in the anno while other GTF may use gene_type or won't even provide this attribute.

You could make the script more universal with this replacement:

if ( as.logical(opt$mito) && ("gene_biotype" %in% colnames(anno)) ){
  anno$mito <- ucfirst(as.character(tolower(anno$gene_biotype) %in% cleanlist(opt$mito_biotypes) | tolower(anno$chromosome) %in% cleanlist(opt$mito_chr)))
} else if (as.logical(opt$mito)  && ("gene_type" %in% colnames(anno))  ){
  anno$mito <- ucfirst(as.character(tolower(anno$gene_type) %in% cleanlist(opt$mito_biotypes) | tolower(anno$chromosome) %in% cleanlist(opt$mito_chr)))
} else if (as.logical(opt$mito)){
  die('ERROR: gene_biotype or gene_type attributes are missing. Mitochondrial features will not be flagged ')
}

I just did it on my galaxy and it works nicely.
Cheers

[pinin4fjords]

Thanks @cgirardot. I intend for this code to live at https://github.com/ebi-gene-expression-group/atlas-gene-annotation-manipulation in future- I just hadn't got round to switching the Galaxy wrapper to use it.

I think a neater solution will be to parameterise the type field with 'gene_biotype' as the default. Would you be able to raise a PR to that effect at the above? Otherwise I'll get to it when I can.

[cgirardot]

I also thought about your solution but I only few minutes to fix this and went for this hack. I am not gonna have the time to make this better, sorry.

[pcm32]

Thanks for reporting it Charles, I'm glad to see that the tools are serving you guys, we will try to fix this soon and make a new release.