Issue with create_background - found in EWCE

[Al-Murphy]

1. Bug description

create_background when all species are the same as the output species and background is not null. In this case currently the background input is ignored instead of just being returned:

species_list <- c(species1,species2)
    gene_var <- if(as_output_species) "ortholog_gene" else "input_gene"
    if(all(species_list==output_species)){
        #### If all species are the same, just use all_genes ####
        gene_map <- all_genes(species = output_species, 
                              method = method,
                              verbose = verbose)
        bg <- gene_map$Gene.Symbol
        return(bg)
    }

This should be changed to:

species_list <- c(species1,species2)
    gene_var <- if(as_output_species) "ortholog_gene" else "input_gene"
    if(all(species_list==output_species)){
        if(is.null(bg)){
            #### If all species are the same, just use all_genes ####
            gene_map <- all_genes(species = output_species, 
                                  method = method,
                                  verbose = verbose)
            bg <- gene_map$Gene.Symbol
        } else {
            bg <- unique(bg) 
        }
        return(bg)
    }

This affects EWCE in prepare_genesize_control_network and bootstrap_enrichment_test as instead of using the user's inputted background list, it is now creating a new one.

I've added code in EWCE to avoid this for now but it would be better to just update at orthogene

[bschilder]

Great catch, just added this. Thanks @Al-Murphy!

Will push as soon as i get past some annoying documentation issues that popped up:

> devtools::document()
ℹ Updating orthogene documentation
ℹ Loading orthogene
 Error: 'gprofiler_orgs' is not an exported object from 'namespace:orthogene'

[bschilder]

@Al-Murphy I noticed in MSS you both store sumstatsColHeaders as internal data and as a .rda file (essentially storing it in multiple locations, internal and external). Is there a reason you need to do both? My understanding was that you only have to do one or the other.

#' usethis::use_data(sumstatsColHeaders,overwrite = TRUE, internal=TRUE)
#' save(sumstatsColHeaders,
#'       file="data/sumstatsColHeaders.rda")

https://github.com/neurogenomics/MungeSumstats/blob/118e7f4cd94aa7b4c3af1b592aa811587200582a/R/data.R#L37

[Al-Murphy]

Ah took me a while to remember, I store as an rda so users can more easily access it since it's meant to be editable for them! But yeah no reason other than that to do both (of course they could access it the other way but I just tried to make it as easy as possible)

[bschilder]

Ok, gotcha!

It turns out my issue was that in the data.R file I was specifying the name of the variable instead of NULL. This meant that devtools expected an exported object of that name. I also had to manually add the @name tag, as well as the @usage tag which isn't applicable to internal objects apparently:

before

....
#' @usage orthogene:::gprofiler_orgs
"gprofiler_orgs"

after

.... 
#' @name gprofiler_orgs
NULL