GSE add links

add_links.R

library(readxl)
library(writexl)
library(gemma.R)
library(GEOquery)
library(dplyr)
library(glue)

data = readxl::read_excel('~/Downloads/Bad_QC_datasets_together.xlsx')

datasets = data$dataset

# pubmed ids ------------
pubmed_ids = datasets %>% sapply(function(name){
    name = gsub('\\..*?$','',name)
    geo = GEOquery::getGEO(name)
    
    geo %>% lapply(\(x){
        attributes(attributes(x)$experimentData)$other$pubmed_id
    }) %>% unique %>% unlist %>% paste0(collapse = ', ') %>% 
        gsub('\n',', ',.)
})

# gemma links ------------
set_gemma_user('OganM',readLines('~/gemma.R/auth'))
gemma_data = get_datasets_by_ids(datasets) %>% get_all_pages()

gemma_links = 
    glue::glue("https://gemma.msl.ubc.ca/expressionExperiment/showExpressionExperiment.html?id={gemma_data$experiment.ID}") %>% 
    {.[match(datasets,gemma_data$experiment.shortName)]}


# geo pages--------

geo_links = glue::glue("https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc={gsub('\\\\..*?$','',datasets)}")


# multiqc report-----------

multiqc_links = glue::glue("https://luigi.pavlab.msl.ubc.ca/rnaseq-pipeline/experiment/{datasets}/report")



data$pubmed_ids = pubmed_ids
data$gemma_links = writexl::xl_hyperlink(gemma_links)
data$geo_links = writexl::xl_hyperlink(geo_links)
data$multiqc_links = writexl::xl_hyperlink(multiqc_links)

writexl::write_xlsx(data,'bad_data_links.xlsx')