yeast download for jesse

differential download.r

library(gemmaAPI)

# identify all yeast datasets
yeastStudies  = taxonInfo('yeast',request = 'datasets',limit = 0)

studyIDs = yeastStudies %>% purrr::map_chr('id')

# get metadata for yeast studies
yeastMetadata = studyIDs %>% lapply(compileMetadata,outputType = 'list') 

quality = yeastMetadata %>% purrr::map('experimentData') %>% purrr::map_dbl('geeq.qualityScore')
quality %>% hist()
# count datasets for platforms. note that there are only 3 experiments between
# the most common (GPL2529, 58 experiments, 4618 annotated genes, 10928 probes)
# and the second most common (GPL90, 55 experiments, 5549 annotated genes, 9335
# probes). don't have much experience with yeast platforms myself so can't
# comment on their relative quality. This gets the data for both of these,
# saved to different directories

platforms = yeastMetadata %>% 
    purrr::map('experimentData') %>% 
    purrr::map_chr('platformName')

popularPlatforms =  platforms %>% 
    table %>% 
    sort(decreasing = TRUE)  %>%
    names %>% {.[1:2]}


# filter metadata do only include the datasets with the popularPlatforms
yeastMetadata = yeastMetadata[platforms %in% popularPlatforms]


# get platforms for each dataset
platforms =  yeastMetadata %>% 
    purrr::map('experimentData') %>% 
    purrr::map_chr('platformName')

studyIDs = yeastMetadata %>% 
    purrr::map('experimentData') %>% purrr::map_chr('datasetID')

diffs = studyIDs %>% 
    lapply(datasetInfo,request = 'differential', offset =0, limit = 0) 

differentials = diffs %>% purrr::map(names)

# split the platforms in case you want to analyze them separately or ignore one of them
popularPlatforms %>% lapply(function(platform){
    studyIDs = yeastMetadata[platforms == platform] %>% 
        purrr::map('experimentData') %>% purrr::map_chr('datasetID')
    
    # get differential IDs from the datasets
    differentials = studyIDs %>% 
        lapply(datasetInfo,request = 'differential', offset =0, limit = 0) %>%
        purrr::map(names)
    
    # download each differential for each dataset
    dir.create(platform, showWarnings = FALSE)
    names(differentials) %>% lapply(function(x){

        differentials[[x]] %>% lapply(function(id){
            print(x)
            # these two studies returned errors.
            if(x %in% c('GSE17716','GSE40399')){
                return(NULL)
            }
            # if is safe to remove 
            if(!file.exists(glue::glue('{platform}/{x}_{id}'))){
                datasetInfo(x, request = 'degs',differential = id,file = glue::glue('{platform}/{x}_{id}'),return=FALSE)

            } else{
                print('exists already. skipping')
            }
            
        })
        
    })
})