The package works fine, but it seems not to provide access to fields, but only to the whole data in a certain file (function hdf5load(file, load = TRUE, verbosity = 0, tidy = FALSE)) Since hdf5 organizes the data and metadata in a hierarchical structure, we must explore in our problem (file>7GB), but it seems not possible with the functions in the package. Any suggestions? Please, let me know about such capabilities or intentions to have them available in the future.
On Thu, 19 Mar 2009, Dani wrote:> The package works fine, but it seems not to provide access to fields, > but only to the whole data in a certain file (function hdf5load(file, > load = TRUE, verbosity = 0, tidy = FALSE)) > > Since hdf5 organizes the data and metadata in a hierarchical > structure, we must explore in our problem (file>7GB), but it seems not > possible with the functions in the package. > > Any suggestions? Please, let me know about such capabilities or > intentions to have them available in the future.You might also want to try the bioconductor mailing list -- they have shown interest in HDF in the past. -thomas Thomas Lumley Assoc. Professor, Biostatistics tlumley at u.washington.edu University of Washington, Seattle
Hi Thomas, Thanks for responding. You are right about bioconductor, however their solution was to implement RHDF5 package for linux only, which allows selecting fields, but not ranges of the data. Any idea about ranges of data? My datafiles are as big as 20GB. thx, Dani On Fri, Mar 20, 2009 at 1:19 AM, Thomas Lumley <tlumley at u.washington.edu> wrote:> On Thu, 19 Mar 2009, Dani wrote: > >> The package works fine, but it seems not to provide access to fields, >> but only to the whole data in a certain file (function hdf5load(file, >> load = TRUE, verbosity = 0, tidy = FALSE)) >> >> Since hdf5 organizes the data and metadata in a hierarchical >> structure, we must explore in our problem (file>7GB), but it seems not >> possible with the functions in the package. >> >> Any suggestions? Please, let me know about such capabilities or >> intentions to have them available in the future. > > You might also want to try the bioconductor mailing list -- they have shown > interest in HDF in the past. > > ? ? ?-thomas > > Thomas Lumley ? ? ? ? ? ? ? ? ? Assoc. Professor, Biostatistics > tlumley at u.washington.edu ? ? ? ?University of Washington, Seattle > > >