Contents

1 Using the DelayedArray infrastructure

A remote dataset is accessed by giving the URL of the server, the type of the server (At present the only valid value is hsds.), the file domain (path to the HDF5 file) and the full path to the dataset inside the HDF5 file.

1.1 Interface to HSDS (HDF Object Store)

if (check_hsds()) {
 da <- HSDSArray(URL_hsds(), 'hsds', 
      '/shared/bioconductor/patelGBMSC.h5', '/assay001')
 da
}
## <65218 x 864> HSDSMatrix object of type "double":
##                [,1]       [,2]       [,3] ...    [,863]    [,864]
##     [1,]  35.650170  41.287100 357.417120   . 123.94820 171.27235
##     [2,]  13.278500   8.313680  21.655700   .  12.10447  12.32039
##     [3,] 162.852500   7.178148  74.442600   .  58.79981 112.98180
##     [4,]  57.402100  26.245630  77.536000   . 122.45777 136.59439
##     [5,]  23.437520 374.456800  36.411020   .  39.67649 217.15135
##      ...          .          .          .   .         .         .
## [65214,]    7.14753    6.25654    5.77305   .  14.51280   8.78161
## [65215,]    0.00000    1.22547    4.00000   .   8.67364   6.26960
## [65216,]    0.00000    6.77233    9.72984   .  13.68860   8.16911
## [65217,]   23.61430    4.41583    7.26575   .   4.55394   9.04190
## [65218,]    0.00000    0.00000    3.28224   .   2.05464   3.00000

Again we have DelayedArray capabilities.

if (check_hsds()) {
 apply(da[,1:4],2,sum)
}
## [1] 1965028 1267166 1627512 1338411