diff --git a/README.html b/README.html index 37fcc26..f76a607 100644 --- a/README.html +++ b/README.html @@ -769,12 +769,22 @@

BPCells backend for DelayedArr

Matrix Storage Format

BPCells provide three format:

    -
  1. Directory of files * writeBPCellsDirArray
    2. -
  2. Hdf5 file
    3. -
  3. in memory * writeBPCellsMemArray
    4. +
  4. Directory of files +
    5. +
  5. Hdf5 file +
    6. +
  6. in memory +
-

"*" means the format has been implemented in BPCellsArray package, followed by the function to implement this format.

-

Matrices can be stored in a directory on disk, in memory, or in an HDF5 file. Saving in a directory on disk is a good default for local analysis, as it provides the best I/O performance and lowest memory usage. The HDF5 format allows saving within existing hdf5 files to group data together, and the in memory format provides the fastest performance in the event memory usage is unimportant.

+

Matrices can be stored in a directory on disk, in memory, or in an HDF5 file. Saving in a directory on disk is a good default for local analysis, as it provides the best I/O performance and lowest memory usage. The HDF5 format allows saving within existing hdf5 files to group data together, and the in memory format provides the fastest performance in the event memory usage is unimportant. So when using as(object, "BPCellsArray") or as(object, "BPCellsMatrix"), the default behavior will be as(object, "BPCellsDirMatrix").

Details see: https://bnprks.github.io/BPCells/articles/web-only/bitpacking-format.html

library(BPCellsArray)
 library(SingleCellExperiment)
@@ -876,7 +886,7 @@ 
Matrix Storage Format

 #> Storage order: column major
 #> 
 #> Queued Operations:
-#> 1. Load compressed matrix from directory /tmp/RtmpRHG9Yx/BPCells2e32ea6b10f2ac
+#> 1. Load compressed matrix from directory /tmp/RtmpG0nbIy/BPCells336a81150062bf

If you do delayed operations with this assay, the class may be changed, that’s because all of BPCells operations are lazy, no real work is performed on the matrix until the result needs to be returned as an R object or written to disk. You can coerce it into a dense matrix or dgCMatrix to get a actual R object.

assay(sce, "counts")[1:10, 1:10]
 #> 10 x 10 DelayedMatrix object with class BPCellsMatrix
@@ -888,7 +898,7 @@ 
Matrix Storage Format

 #> Storage order: column major
 #> 
 #> Queued Operations:
-#> 1. Load compressed matrix from directory /tmp/RtmpRHG9Yx/BPCells2e32ea6b10f2ac
+#> 1. Load compressed matrix from directory /tmp/RtmpG0nbIy/BPCells336a81150062bf
 #> 2. Select rows: 1, 2 ... 10 and cols: 1, 2 ... 10
 as.matrix(assay(sce, "counts")[1:10, 1:10])
 #>           Cell_001 Cell_002 Cell_003 Cell_004 Cell_005 Cell_006 Cell_007
@@ -942,7 +952,7 @@ 
Matrix Storage Format

 #> Storage order: column major
 #> 
 #> Queued Operations:
-#> 1. Load compressed matrix from directory /tmp/RtmpRHG9Yx/BPCells2e32ea6b10f2ac
+#> 1. Load compressed matrix from directory /tmp/RtmpG0nbIy/BPCells336a81150062bf
 #> 2. Scale columns by 0.984, 1.05 ... 1
 #> 3. Transform log1p
 #> 4. Scale by 1.44
diff --git a/README.md b/README.md index b79d67a..3a9c6c2 100644 --- a/README.md +++ b/README.md @@ -55,19 +55,23 @@ Other non-lazied operations: BPCells provide three format: -1. Directory of files \* `writeBPCellsDirArray` +1. Directory of files + - read: `readBPCellsDirMatrix` + - write: `writeBPCellsDirArray` 2. Hdf5 file -3. in memory \* `writeBPCellsMemArray` - -"\*" means the format has been implemented in `BPCellsArray` package, -followed by the function to implement this format. + - read: `readBPCellsHDF5Matrix` + - write: `writeBPCellsHDF5Array` +3. in memory + - write: `writeBPCellsMemArray` Matrices can be stored in a directory on disk, in memory, or in an HDF5 file. Saving in a directory on disk is a good default for local analysis, as it provides the best I/O performance and lowest memory usage. The HDF5 format allows saving within existing hdf5 files to group data together, and the in memory format provides the fastest performance -in the event memory usage is unimportant. +in the event memory usage is unimportant. So when using `as(object, +"BPCellsArray")` or `as(object, "BPCellsMatrix")`, the default behavior +will be `as(object, "BPCellsDirMatrix")`. Details see: @@ -190,7 +194,7 @@ assay(sce, "counts") #> Storage order: column major #> #> Queued Operations: -#> 1. Load compressed matrix from directory /tmp/RtmpRHG9Yx/BPCells2e32ea6b10f2ac +#> 1. Load compressed matrix from directory /tmp/RtmpG0nbIy/BPCells336a81150062bf ``` If you do delayed operations with this assay, the class may be changed, @@ -210,7 +214,7 @@ assay(sce, "counts")[1:10, 1:10] #> Storage order: column major #> #> Queued Operations: -#> 1. Load compressed matrix from directory /tmp/RtmpRHG9Yx/BPCells2e32ea6b10f2ac +#> 1. Load compressed matrix from directory /tmp/RtmpG0nbIy/BPCells336a81150062bf #> 2. Select rows: 1, 2 ... 10 and cols: 1, 2 ... 10 as.matrix(assay(sce, "counts")[1:10, 1:10]) #> Cell_001 Cell_002 Cell_003 Cell_004 Cell_005 Cell_006 Cell_007 @@ -275,7 +279,7 @@ assay(sce, "logcounts") #> Storage order: column major #> #> Queued Operations: -#> 1. Load compressed matrix from directory /tmp/RtmpRHG9Yx/BPCells2e32ea6b10f2ac +#> 1. Load compressed matrix from directory /tmp/RtmpG0nbIy/BPCells336a81150062bf #> 2. Scale columns by 0.984, 1.05 ... 1 #> 3. Transform log1p #> 4. Scale by 1.44 diff --git a/vignettes/BPCellsArray.Rmd b/vignettes/BPCellsArray.Rmd index 26e0b38..af93ea7 100644 --- a/vignettes/BPCellsArray.Rmd +++ b/vignettes/BPCellsArray.Rmd @@ -61,19 +61,23 @@ Other non-lazied operations: ## Matrix Storage Format BPCells provide three format: -1. Directory of files * `writeBPCellsDirArray` -2. Hdf5 file -3. in memory * `writeBPCellsMemArray` - -"*" means the format has been implemented in `BPCellsArray` package, followed by -the function to implement this format. +1. Directory of files + - read: `readBPCellsDirMatrix` + - write: `writeBPCellsDirArray` +2. Hdf5 file + - read: `readBPCellsHDF5Matrix` + - write: `writeBPCellsHDF5Array` +3. in memory + - write: `writeBPCellsMemArray` Matrices can be stored in a directory on disk, in memory, or in an HDF5 file. Saving in a directory on disk is a good default for local analysis, as it provides the best I/O performance and lowest memory usage. The HDF5 format allows saving within existing hdf5 files to group data together, and the in memory format provides the fastest performance in the event memory usage is -unimportant. +unimportant. So when using `as(object, "BPCellsArray")` or `as(object, +"BPCellsMatrix")`, the default behavior will be `as(object, +"BPCellsDirMatrix")`. Details see: