In Bioconductor/GDSArray: Representing GDS files as array-like objects

5/1/2018, summary 1. "extract_array", updated, to return only array. (for 1-dimensional data also). 2. dim, dimnames, type for "GDSArray" works by implementing "extract_array" correctly. 3. show method works for "DelayedSubset" object inside vignette, by depending on S4Vector 0.17.43. 4. gdsnodes(GDSFile) returns only the gds nodes that could construct unique ‘GDSArray’ objects, which means that the gds node has non-zero-dimensions, and is actually array, and all ‘GDSArray’s returned from these nodes are unique (by excluding the gds nodes that has `code~ prefix).

4/23/2018

permute, 3-dimensional data, ??
GDSArray, >2-dimensional, all in variant X sample X other.
GDS file, "data" and "~data" are identical except permuted???
if yes, GDSFile only show "subnodes" of "data", not "~data". -- done!
same for gdsnodes(file). -- to do or not?
Because GDSArray() returns the array in a fixed order: vaiant X sample X other dimension. No need for permutation to return the same data.
extract_array, return array(). called inside: DelayedArray:::.extract_array_with_dimnames(). -- done!
extract_array, method for "GDSArray" instead of "GDSArraySeed"...
https://github.com/Bioconductor/DelayedArray/blob/master/vignettes/02-Implementing_a_backend.Rmd

4/20/2018

gf$genotype$data -- correct!
gf$genotype$~data -- dimnames
gf$sample.id, gf$annotation$qual -- 1-dimensional gds nodes not correct.
GDSFile show method, $completion method, show gdsnodes(GDSFile) instead of gdsnodes(x@file). -- done!
add method type,GDSArray, to be consistent with generic set in DelayedArray. -- done!
add method dim, dimnames, GDSArray. -- done!
add method documentations. -- done!

4/17/2018

implement: ".get_gdsdata_isarray", extract objdesp.gdsn(f,name)$is.array --done!
implement: ".get_gdsdata_non1D_array", extract >=2 dimensional array data
implenentation. -- done!
modify: (makeSEFromGDS:showAvailable()$name -- todo...
modify: ".get_gdsdata_dim",
return NULL for non-array gds nodes. -- done!
return objdesp$dim for all array gds nodes. (.get_gdsdata_isrray == TRUE) -- done!
modify: ".get_gdsdata_dimnames", check ".get_gdsdata_dim". -- done!
if is.null(dim), return NULL (for non-array gds nodes.)
if any(dim == 0), return NULL.
modify: ".read_gdsdata_first_val", check ".get_gdsdata_dim". -- done!
if is.null(dim), return NULL (for non-array data, e.g., Folder, VFolder...).
if any(dim == 0), return NULL (for 0-dim nodes, e.g., file, "phase/extra", "phase/extra.index", "genotype/extra"...)
modify: "GDSArraySeed()" constructor. check ".get_gdsdata_dim". -- done!
if is.null(dim), wmsg("Not a valid gds node for constructing GDSArray")
if any(dim == 0), wmsg("The dimension of gds node is: ", ", which cound not construct GDSArray")
modify "GDSFile$" method, -- done!
"gdsnodes(file.gds)", returns all nodes, including folder nodes.
"gdsnodes(GDSFile)", return gdsnodes that are array and !any(dim==0).

gf$variant.id <1348> GDSArray object of type "integer": Error in dimnames(x) <- value : 'dimnames' applied to non-array gf$genotype$data -- correct! gf$phase$data -- correct!

3/13/2018