GCSFilesystem provides an unified interface for mounting Google Cloud Storage buckets to your local system. After a bucket has been mounted, you can view and access the files and folders in the bucket using your file browser as if they are stored locally. You must has GCSDokan on Windows or gcsfuse on Linux or macOs prior to using this package. You can find documents on how to install the dependencies by clicking the links above.
The package uses Google application default credentials to authenticate with Google. In most case, you need to provide a service account JSON file to verify your identity. For automatically finding your credentials when using the package, the path to the JSON file can be stored at the environment variable GOOGLE_APPLICATION_CREDENTIALS. You can also manually provide the JSON file to the package by specifying the argument key_file to the function gcs_mount. The document of how to create a service account can be found here.
Note that there are some buckets which allow anonymous access. For windows, these buckets can be directly mounted without a credentials. For Linux and macOs, the current version of gcsfuse does not support anonymous access and you still need to provide a credentials file to access the buckets.
You can use gcs_mount to mount a bucket on your machine. In the example, we will mount the bucket genomics-public-data to a temporary directory in R
remote_bucket <- "genomics-public-data"
temp_dir <- paste0(tempdir(),"/GCSFilesystemExample")
gcs_mount(remote_bucket, temp_dir)Note that the function can also be used to mount a directory inside a bucket. For example, you can use gcs_mount("genomics-public-data/clinvar", temp_dir) to mount the folder clinvar to your temporary directory. After mounting the package, you can browse the files in your file explore. Here we can list all files in R using list.files
list.files(temp_dir)
#> [1] "1000-genomes" "1000-genomes-phase-3"
#> [3] "clinvar" "cwl-examples"
#> [5] "ftp-trace.ncbi.nih.gov" "gatk-examples"
#> [7] "linkage-disequilibrium" "NA12878.chr20.sample.bam"
#> [9] "NA12878.chr20.sample.DeepVariant-0.7.2.vcf" "platinum-genomes"
#> [11] "precision-fda" "README"
#> [13] "references" "resources"
#> [15] "simons-genome-diversity-project" "test-data"
#> [17] "ucsc"You can find all mount points by gcs_list_mountpoints
gcs_list_mountpoints()
#> remote mountpoint
#> 1 genomics-public-data C:/Users/wangj/AppData/Local/Temp/Rtmpqelnec/GCSFilesystemExampleFinally, after using the bucket, you can unmount it via gcs_unmount
Some buckets have billing project enabled, which means you are responsible for all the charges that occurs when accessing the bucket. For avoiding unintentional cost, the billing project is not enabled by default. If you want to access this type of buckets, you must specify your project Id in the argument billing when calling gcs_mount(e.g. gcs_mount("bucket", "mount-point", billing = "my-project-Id")). Otherwise, you would not be able to see the files in the bucket. Please note that it is not recommended to add the argument billing for all the call to gcs_mount. If you did it, you WILL be changed by Google even if you are trying to access a bucket without billing project enabled.
Since accessing remote files are relatively expensive, the information of files and folders in the mounted bucket will be cached for a certain period of time. The changes to the remote bucket will not be immediately visible until the local information has been expired. By default, the refresh rate is 60 seconds. You can change the refresh rate via the argument refresh when mounting a bucket.
Certain optimization can be made to facilitate your access to the Google Bucket. Since accessing remote files has much higher latency than accessing local files, using local cache can greatly reduce the number of remote requests and reuse the data that has been downloaded before. The cache will be enabled by default and the cache data will be stored on disk. Only Windows users are allowed to change the cache setting, the available cache types are none, disk and memory. They can be changed via the argument cache_type and cache_arg in the function gcs_mount.