diff --git a/.Rbuildignore b/.Rbuildignore
index 91114bf..6ad9034 100644
--- a/.Rbuildignore
+++ b/.Rbuildignore
@@ -1,2 +1,3 @@
 ^.*\.Rproj$
 ^\.Rproj\.user$
+^.*\.h5Seurat$
diff --git a/.gitignore b/.gitignore
index 5b6a065..c4cfafb 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,3 +2,4 @@
 .Rhistory
 .RData
 .Ruserdata
+*.h5Seurat
diff --git a/DESCRIPTION b/DESCRIPTION
index 5a1f2bf..8ba42a6 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: SeuratData
 Type: Package
 Title: Install and Manage Seurat Datasets
-Version: 0.1.0
+Version: 0.1.0.9001
 Date: 2019-07-17
 Authors@R: c(
   person(given = 'Rahul', family = 'Satija', email = 'rsatija@nygenome.org', role = 'aut', comment = c(ORCID = '0000-0001-9448-8833')),
@@ -21,8 +21,15 @@ RoxygenNote: 6.1.1
 Imports:
     cli,
     crayon,
+    methods,
+    Seurat (>= 3.0.2),
+    rappdirs,
     stats,
+    tools,
     utils
-Collate: 
+Collate:
     'zzz.R'
+    'h5seurat.R'
     'seurat_data.R'
+Suggests:
+    hdf5r (>= 1.2.0)
diff --git a/NAMESPACE b/NAMESPACE
index bd01537..42071e5 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -1,10 +1,52 @@
 # Generated by roxygen2: do not edit by hand
 
+S3method(AppendData,H5File)
+S3method(AppendData,character)
+S3method(IndexH5Seurat,H5File)
+S3method(IndexH5Seurat,character)
+S3method(LoadH5Seurat,H5File)
+S3method(LoadH5Seurat,character)
+S3method(SaveH5Seurat,Seurat)
+S3method(SaveH5Seurat,default)
+S3method(as.list,H5Group)
+S3method(print,h5SI)
+export(AppendData)
 export(AvailableData)
+export(IndexH5Seurat)
 export(InstallData)
 export(InstalledData)
+export(LoadData)
+export(LoadH5Seurat)
 export(RemoveData)
+export(SaveH5Seurat)
 export(UpdateData)
+importFrom(Seurat,"Idents<-")
+importFrom(Seurat,"Key<-")
+importFrom(Seurat,"Misc<-")
+importFrom(Seurat,"Project<-")
+importFrom(Seurat,"VariableFeatures<-")
+importFrom(Seurat,AddMetaData)
+importFrom(Seurat,Assays)
+importFrom(Seurat,Command)
+importFrom(Seurat,CreateAssayObject)
+importFrom(Seurat,CreateDimReducObject)
+importFrom(Seurat,DefaultAssay)
+importFrom(Seurat,Embeddings)
+importFrom(Seurat,GetAssayData)
+importFrom(Seurat,Idents)
+importFrom(Seurat,JS)
+importFrom(Seurat,Key)
+importFrom(Seurat,Loadings)
+importFrom(Seurat,Misc)
+importFrom(Seurat,Project)
+importFrom(Seurat,Reductions)
+importFrom(Seurat,SetAssayData)
+importFrom(Seurat,Stdev)
+importFrom(Seurat,Tool)
+importFrom(Seurat,VariableFeatures)
+importFrom(Seurat,as.Graph)
+importFrom(Seurat,as.Seurat)
+importFrom(Seurat,as.sparse)
 importFrom(cli,rule)
 importFrom(cli,symbol)
 importFrom(crayon,blue)
@@ -14,8 +56,15 @@ importFrom(crayon,col_nchar)
 importFrom(crayon,green)
 importFrom(crayon,red)
 importFrom(crayon,yellow)
+importFrom(methods,"slot<-")
+importFrom(methods,new)
+importFrom(methods,slot)
+importFrom(methods,slotNames)
+importFrom(rappdirs,user_data_dir)
 importFrom(stats,na.omit)
+importFrom(tools,file_ext)
 importFrom(utils,available.packages)
+importFrom(utils,data)
 importFrom(utils,install.packages)
 importFrom(utils,packageVersion)
 importFrom(utils,remove.packages)
diff --git a/R/h5seurat.R b/R/h5seurat.R
new file mode 100644
index 0000000..8946261
--- /dev/null
+++ b/R/h5seurat.R
@@ -0,0 +1,1529 @@
+#' @include zzz.R
+#'
+NULL
+
+#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+# Generics
+#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+#' Append data from an h5Seurat file to a preexisting \code{Seurat} object
+#'
+#' @inheritParams LoadH5Seurat
+#' @param object A \code{Seurat} object to append data to
+#'
+#' @return A \code{Seurat} object with the extra data requested
+#'
+#' @export
+#'
+#' @seealso \code{\link{LoadH5Seurat}} \code{\link{LoadData}}
+#'
+#' @keywords internal
+#'
+AppendData <- function(file, object, ...) {
+  if (!requireNamespace('hdf5r', quietly = TRUE)) {
+    stop("Please install hdf5r to enable appending data to a Seurat object from an h5Seurat file", call. = FALSE)
+  }
+  UseMethod(generic = 'AppendData', object = file)
+}
+
+#' Index an H5Seurat object
+#'
+#' @inheritParams LoadH5Seurat
+#'
+#' @return An object of class \code{h5SI}
+#'
+#' @rdname IndexH5Seurat
+#'
+#' @export
+#'
+#' @seealso \code{\link{LoadH5Seurat}} \code{\link{SaveH5Seurat}}
+#'
+#' @keywords internal
+#'
+IndexH5Seurat <- function(file, ...) {
+  if (!requireNamespace('hdf5r', quietly = TRUE)) {
+    stop("Please install hdf5r to enable indexing h5Seurat files", call. = FALSE)
+  }
+  UseMethod(generic = 'IndexH5Seurat', object = file)
+}
+
+#' Load a saved \code{Seurat} object from an h5Seurat file
+#'
+#' @param file Name of h5Seurat or connected h5Seurat file to load
+#' @param assays One of:
+#' \itemize{
+#'   \item A character vector with names of assays
+#'   \item A character vector with one or more of \code{counts}, \code{data},
+#'   \code{scale.data} describing which slots of \strong{all assays} to load
+#'   \item A named list where each entry is either the name of an assay or a vector
+#'   describing which slots (described above) to take from which assay
+#'   \item \code{NULL} for all assays
+#' }
+#' @param reductions One of:
+#' \itemize{
+#'   \item A character vector with names of reductions
+#'   \item \code{NULL} for all reductions
+#'   \item \code{NA} or \code{FALSE} for no reductions
+#' }
+#' \strong{Note}: Only reductions associated with an assay loaded in \code{assays}
+#' @param graphs One of:
+#' \itemize{
+#'   \item A character vector with names of graphs
+#'   \item \code{NULL} for all graphs
+#'   \item \code{NA} or \code{FALSE} for no reductions
+#' }
+#' @param meta.data Load object metadata?
+#' @param commands Load command information
+#' @param misc Load miscellaneous data?
+#' @param tools Load tool-specific information?
+#' @param verbose Show progress updates
+#' @param ... Arguments passed to other methods
+#'
+#' @return If \code{type} is info, information about the data contained within the
+#' h5Seurat file. Otherwise, a \code{Seurat} object with the data requested
+#'
+#' @export
+#'
+#' @seealso \code{\link{SaveH5Seurat}} \code{\link{IndexH5Seurat}}
+#'
+#' @keywords internal
+#'
+LoadH5Seurat <- function(file, ...) {
+  if (!requireNamespace('hdf5r', quietly = TRUE)) {
+    stop("Please install hdf5r to enable loading Seurat objects from h5Seurat files", call. = FALSE)
+  }
+  UseMethod(generic = 'LoadH5Seurat', object = file)
+}
+
+#' Save a \code{Seurat} object to a h5Seurat file
+#'
+#' @param object A \code{Seurat} object
+#' @param filename Name for h5Seurat file
+#' @param overwrite If an h5Seurat file of name \code{filename} exists, overwrite it?
+#' @param verbose Show progress updates
+#' @param ... Arguments passed to other methods
+#'
+#' @return Invisibly returns the name of the h5Seurat file
+#'
+#' @export
+#'
+#' @seealso \code{\link{LoadH5Seurat}} \code{\link{IndexH5Seurat}}
+#'
+#' @keywords internal
+#'
+SaveH5Seurat <- function(object, filename, overwrite = FALSE, verbose = TRUE, ...) {
+  if (!requireNamespace('hdf5r', quietly = TRUE)) {
+    stop("Please install hdf5r to enable saving Seurat object to h5Seurat files", call. = FALSE)
+  }
+  UseMethod(generic = 'SaveH5Seurat', object = object)
+}
+
+#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+# S3 Methods
+#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+#' @rdname AppendData
+#' @method AppendData character
+#' @export
+#'
+AppendData.character <- function(
+  file,
+  object,
+  assays = NULL,
+  reductions = NULL,
+  graphs = NULL,
+  overwrite = FALSE,
+  verbose = TRUE,
+  ...
+) {
+  if (!file.exists(file)) {
+    stop("Cannot find h5Seurat file ", file, call. = FALSE)
+  }
+  hfile <- hdf5r::H5File$new(filename = file, mode = 'r')
+  on.exit(expr = hfile$close_all())
+  return(AppendData(
+    file = hfile,
+    object = object,
+    assays = assays,
+    reductions = reductions,
+    graphs = graphs,
+    overwrite = overwrite,
+    verbose = verbose,
+    ...
+  ))
+}
+
+#' @importFrom methods slot slot<-
+#' @importFrom Seurat Assays Reductions Command
+#'
+#' @rdname AppendData
+#' @method AppendData H5File
+#' @export
+#'
+AppendData.H5File <- function(
+  file,
+  object,
+  assays = NULL,
+  reductions = NULL,
+  graphs = NULL,
+  overwrite = FALSE,
+  verbose = TRUE,
+  ...
+) {
+  index <- IndexH5Seurat(file = file)
+  cells <- file[['cell.names']][]
+  if (!all(cells == colnames(x = object))) {
+    stop("Mismatched cells between the h5Seurat file and the Seurat object", call. = FALSE)
+  }
+  # Load assays
+  assays.all <- is.null(x = assays)
+  if (is.null(x = assays)) {
+    assays <- names(x = index$assays)
+  }
+  assays <- GetAssays(assays = assays, index = index)
+  if (!overwrite) {
+    assays <- assays[!names(x = assays) %in% Assays(object = object)]
+  } else if (verbose) {
+    overwritten <- names(x = assays)[names(x = assays) %in% Assays(object = object)]
+    if (length(x = overwritten) > 0) {
+      message(
+        "Overwriting the following assays: ",
+        paste(overwritten, collapse = ', ')
+      )
+    }
+  }
+  if (length(x = assays) < 1) {
+    warning("All assay objects loaded", call. = FALSE, immediate. = TRUE)
+  } else {
+    if (verbose) {
+      message("Adding data for ", length(x = assays), " assays")
+    }
+    for (assay in names(x = assays)) {
+      object[[assay]] <- AssembleAssay(
+        assay = assay,
+        file = file,
+        cells = cells,
+        slots = assays[[assay]],
+        verbose = verbose
+      )
+    }
+  }
+  # Load DimReducs
+  reducs.all <- is.null(x = reductions)
+  if (is.null(x = reductions)) {
+    reductions <- names(x = index$reductions)
+  }
+  if (!(isFALSE(x = reductions) || is.na(x = reductions))) {
+    reductions <- GetDimReducs(
+      reductions = reductions,
+      index = index,
+      assays = Assays(object = object)
+    )
+    if (!overwrite) {
+      reductions <- reductions[!reductions %in% Reductions(object = object)]
+    } else if (verbose) {
+      overwritten <- reductions[reductions %in% Reductions(object = object)]
+      if (length(x = overwritten) > 0) {
+        message(
+          "Overwriting the following dimensional reduction information: ",
+          paste(overwritten, collapse = ', ')
+        )
+      }
+    }
+    if (length(x = reductions) < 1) {
+      warning(
+        "None of the reductions specified are associated with a loaded assay",
+        call. = FALSE,
+        immediate. = TRUE
+      )
+    } else {
+      if (verbose) {
+        message(
+          "Adding dimensional reduction information for ",
+          length(x = reductions),
+          " reductions"
+        )
+      }
+      for (reduc in reductions) {
+        object[[reduc]] <- AssembleDimReduc(
+          reduction = reduc,
+          file = file,
+          cells = cells,
+          default.assay = index$reductions[[reduc]]$assay,
+          embeddings = index$reductions[[reduc]]$cell.embeddings,
+          loadings = index$reductions[[reduc]]$feature.loadings,
+          projected = index$reductions[[reduc]]$feature.loadings.projected,
+          jackstraw = index$reductions[[reduc]]$jackstraw,
+          verbose = verbose
+        )
+      }
+    }
+  } else if (verbose) {
+    message("No dimensional reduction information requested")
+  }
+  # Load Graphs
+  graphs.all <- is.null(x = graphs)
+  if (is.null(x = graphs)) {
+    graphs <- GetGraphs(index = index, assays = Assays(object = object))
+    object.graphs <- Filter(
+      f = function(x) {
+        return(inherits(x = object[[x]], what = 'Graph'))
+      },
+      x = names(x = object)
+    )
+    if (!overwrite) {
+      graphs <- graphs[!graphs %in% object.graphs]
+    } else if (verbose) {
+      overwritten <- graphs[graphs %in% object.graphs]
+      if (length(x = overwritten) > 0) {
+        message(
+          "Overwritting the following graphs: ",
+          paste(overwritten, collapse = ', ')
+        )
+      }
+    }
+    if (length(x = graphs) == 0) {
+      graphs <- FALSE
+    }
+  }
+  if (!(isFALSE(x = graphs) || is.na(x = graphs))) {
+    if (verbose) {
+      message("Adding ", length(x = graphs), " nearest neighbor graphs")
+    }
+    for (graph in graphs) {
+      if (verbose) {
+        message("Adding graph ", graph)
+      }
+      graph.matrix <- as.sparse(x = file[['graphs']][[graph]])
+      rownames(x = graph.matrix) <- colnames(x = graph.matrix) <- cells
+      object[[graph]] <- as.Graph(x = graph.matrix)
+    }
+  } else if (verbose) {
+    message("No nearest neighbor graphs requested")
+  }
+  # Load SeuratCommands
+  if (all(assays.all, reducs.all, graphs.all)) {
+    commands <- names(x = file[['commands']])
+    if (!overwrite) {
+      commands <- setdiff(x = commands, y = Command(object = object))
+    } else if (verbose) {
+      overwritten <- intersect(x = commands, y = Command(object = object))
+      if (length(x = overwritten) > 0) {
+        message(
+          "Overwritting the following command logs: ",
+          paste(overwritten, collapse = ', ')
+        )
+      }
+    }
+    object.commands <- slot(object = object, name = 'commands')
+    if (verbose && length(x = commands) > 0) {
+      message("Adding ", length(x = commands), " command logs")
+    }
+    for (cmd in commands) {
+      object.commands[[cmd]] <- AssembleSeuratCommand(
+        cmd = cmd,
+        file = file,
+        verbose = verbose
+      )
+    }
+  }
+  return(object)
+}
+
+#' Read an HDF5 group into a ist
+#'
+#' @param x An \code{H5Group} object (from hdf5r)
+#' @param ... Ignored
+#'
+#' @return A list with the data contained within \code{x}; if the HDF5 attribute
+#' \code{s4class} is set and is a class, will return an object of class \code{s4class}
+#' instead
+#'
+#' @aliases as.list
+#'
+#' @rdname as.list.H5Group
+#' @method as.list H5Group
+#' @export
+#'
+#' @keywords internal
+#'
+as.list.H5Group <- function(x, ...) {
+  objs <- names(x = x)
+  to.return <- vector(mode = 'list', length = length(x = objs))
+  names(x = to.return) <- objs
+  for (i in objs) {
+    to.return[[i]] <- if (inherits(x = x[[i]], what = 'H5Group')) {
+      as.list(x = x[[i]])
+    } else if (length(x = x[[i]]$dims) > 1) {
+      x[[i]][, ]
+    } else {
+      x[[i]][]
+    }
+  }
+  if (!is.null(x = hdf5r::h5attributes(x = x)$s4class)) {
+    to.return <- tryCatch(
+      expr = do.call(
+        what = 'new',
+        args = c(Class = hdf5r::h5attr(x = x, which = 's4class'), to.return)
+      ),
+      error = function(...) {
+        return(to.return)
+      }
+    )
+  }
+  return(to.return)
+}
+
+#' @rdname IndexH5Seurat
+#' @method IndexH5Seurat character
+#' @export
+#'
+#' @keywords internal
+#'
+IndexH5Seurat.character <- function(file, ...) {
+  if (!file.exists(file)) {
+    stop("Cannot find h5Seurat file ", file, call = FALSE)
+  }
+  hfile <- hdf5r::H5File$new(filename = file, mode = 'r')
+  on.exit(expr = hfile$close_all())
+  return(IndexH5Seurat(file = hfile, ...))
+}
+
+#' @rdname IndexH5Seurat
+#' @method IndexH5Seurat H5File
+#' @export
+#'
+#' @keywords internal
+#'
+IndexH5Seurat.H5File <- function(file, ...) {
+  index <- list()
+  # Get assay information
+  assays <- names(x = file[['assays']])
+  assays <- sapply(
+    X = assays,
+    FUN = function(x) {
+      slots <- c('counts', 'data', 'scale.data')
+      check <- slots %in% names(x = file[['assays']][[x]])
+      names(x = check) <- slots
+      check[['scale.data']] <- check[['scale.data']] && file[['assays']][[x]]$exists(name = 'scaled.features')
+      return(check)
+    },
+    simplify = FALSE,
+    USE.NAMES = TRUE
+  )
+  if (!all(vapply(X = assays, FUN = '[[', FUN.VALUE = logical(length = 1L), 'data'))) {
+    stop("All assays need a 'data' slot", call. = FALSE)
+  }
+  index[['assays']] <- assays
+  # Get dimensional reduction information
+  reductions <- names(x = file[['reductions']])
+  reductions <- sapply(
+    X = reductions,
+    FUN = function(x) {
+      reduc <- file[['reductions']][[x]]
+      slots <- c(
+        'cell.embeddings',
+        'feature.loadings',
+        'feature.loadings.projected',
+        'jackstraw'
+      )
+      check <- slots %in% names(x = reduc)
+      names(x = check) <- slots
+      check[['feature.loadings']] <- check[['feature.loadings']] && reduc$exists(name = 'features')
+      check[['feature.loadings.projected']] <- check[['feature.loadings.projected']] && reduc$exists(name = 'projected.features')
+      check <- c(
+        'assay' = hdf5r::h5attr(
+          x = file[['reductions']][[x]],
+          which = 'active.assay'
+        ),
+        as.list(x = check)
+      )
+      return(check)
+    },
+    simplify = FALSE,
+    USE.NAMES = TRUE
+  )
+  index[['reductions']] <- reductions
+  # Get graph information
+  graphs <- names(x = file[['graphs']])
+  graph.assays <- sapply(X = strsplit(x = graphs, split = '_'), FUN = '[[', 1)
+  graphs <- sapply(
+    X = unique(x = graph.assays),
+    FUN = function(x) {
+      return(grep(pattern = paste0('^', x, '_'), x = graphs, value = TRUE))
+    },
+    simplify = FALSE,
+    USE.NAMES = TRUE
+  )
+  index[['graphs']] <- graphs
+  # Get active.assay
+  index[['active.assay']] <- hdf5r::h5attr(x = file, which = 'active.assay')
+  # Get commands
+  # Get metadata
+  # Get misc
+  # Get tools
+  # Return the index
+  return(ValidateH5SI(x = structure(.Data = index, class = 'h5SI')))
+}
+
+#' @rdname LoadH5Seurat
+#' @method LoadH5Seurat character
+#' @export
+#'
+#' @keywords internal
+#'
+LoadH5Seurat.character <- function(
+  file,
+  type = c('info', 'object'),
+  assays = NULL,
+  reductions = NULL,
+  graphs = NULL,
+  verbose = TRUE,
+  ...
+) {
+  if (!file.exists(file)) {
+    stop("Cannot find h5Seurat file ", file, call. = FALSE)
+  }
+  hfile <- hdf5r::H5File$new(filename = file, mode = 'r')
+  on.exit(expr = hfile$close_all())
+  return(LoadH5Seurat(
+    file = hfile,
+    type = type,
+    assays = assays,
+    reductions = reductions,
+    graphs = graphs,
+    verbose = verbose,
+    ...
+  ))
+}
+
+#' @importFrom methods new slot<-
+#' @importFrom Seurat as.sparse AddMetaData CreateDimReducObject as.Graph
+#' Idents<- Project<- Project Misc<-
+#'
+#' @rdname LoadH5Seurat
+#' @method LoadH5Seurat H5File
+#' @export
+#'
+#' @keywords internal
+#'
+LoadH5Seurat.H5File <- function(
+  file,
+  type = c('info', 'object'),
+  assays = NULL,
+  reductions = NULL,
+  graphs = NULL,
+  verbose = TRUE,
+  ...
+) {
+  type <- match.arg(arg = tolower(x = type), choices = c('info', 'object'))
+  cells <- file[['cell.names']][]
+  index <- IndexH5Seurat(file = file)
+  if (type == 'info') {
+    return(index)
+  }
+  # Load Assays
+  assays.all <- is.null(x = assays)
+  if (is.null(x = assays)) {
+    assays <- names(x = index$assays)
+  }
+  assays <- GetAssays(assays = assays, index = index)
+  if (!index$active.assay %in% names(x = assays)) {
+    warning(
+      "Default assay not requested, using ",
+      names(x = assays)[1],
+      " as default instead",
+      call. = FALSE,
+      immediate. = TRUE
+    )
+    active.assay <- names(x = assays)[1]
+  } else {
+    active.assay <- index$active.assay
+  }
+  assay.objects <- vector(mode = 'list', length = length(x = assays))
+  names(x = assay.objects) <- names(x = assays)
+  for (assay in names(x = assays)) {
+    assay.objects[[assay]] <- AssembleAssay(
+      assay = assay,
+      file = file,
+      cells = cells,
+      slots = assays[[assay]],
+      verbose = verbose
+    )
+  }
+  default.assays <- list(assay.objects[[active.assay]])
+  names(x = default.assays) <- active.assay
+  object <- new(
+    Class = 'Seurat',
+    assays = default.assays,
+    active.assay = active.assay,
+    meta.data = data.frame(row.names = cells),
+    version = package_version(x = hdf5r::h5attr(x = file, which = 'version'))
+  )
+  for (assay in names(x = assay.objects)) {
+    if (assay != active.assay) {
+      object[[assay]] <- assay.objects[[assay]]
+    }
+  }
+  # Load DimReducs
+  reducs.all <- is.null(x = reductions)
+  if (is.null(x = reductions)) {
+    reductions <- names(x = index$reductions)
+  }
+  if (!(isFALSE(x = reductions) || is.na(x = reductions))) {
+    reductions <- GetDimReducs(
+      reductions = reductions,
+      index = index,
+      assays = names(x = assays)
+    )
+    if (length(x = reductions) < 1) {
+      warning(
+        "None of the reductions specified are associated with a loaded assay",
+        call. = FALSE,
+        immediate. = TRUE
+      )
+    } else {
+      for (reduc in reductions) {
+        object[[reduc]] <- AssembleDimReduc(
+          reduction = reduc,
+          file = file,
+          cells = cells,
+          default.assay = index$reductions[[reduc]]$assay,
+          embeddings = index$reductions[[reduc]]$cell.embeddings,
+          loadings = index$reductions[[reduc]]$feature.loadings,
+          projected = index$reductions[[reduc]]$feature.loadings.projected,
+          jackstraw = index$reductions[[reduc]]$jackstraw,
+          verbose = verbose
+        )
+      }
+    }
+  } else if (verbose) {
+    message("No dimensional reduction information requested")
+  }
+  # Load Graphs
+  graphs.all <- is.null(x = graphs)
+  if (is.null(x = graphs)) {
+    graphs <- GetGraphs(index = index, assays = names(x = assays))
+    if (length(x = graphs) == 0) {
+      graphs <- FALSE
+    }
+  }
+  if (!(isFALSE(x = graphs) || is.na(x = graphs))) {
+    for (graph in graphs) {
+      if (verbose) {
+        message("Adding graph ", graph)
+      }
+      graph.matrix <- as.sparse(x = file[['graphs']][[graph]])
+      rownames(x = graph.matrix) <- colnames(x = graph.matrix) <- cells
+      object[[graph]] <- as.Graph(x = graph.matrix)
+    }
+  } else if (verbose) {
+    message("No nearest neighbor graphs requested")
+  }
+  # Load SeuratCommands
+  if (all(assays.all, reducs.all, graphs.all)) {
+    cmds <- vector(mode = 'list', length = length(x = names(x = file[['commands']])))
+    names(x = cmds) <- names(x = file[['commands']])
+    for (cmd in names(x = file[['commands']])) {
+      cmds[[cmd]] <- AssembleSeuratCommand(
+        cmd = cmd,
+        file = file,
+        verbose = verbose
+      )
+    }
+    slot(object = object, name = 'commands') <- cmds
+  }
+  # Load meta.data
+  meta.data <- file[['meta.data']][]
+  rownames(x = meta.data) <- cells
+  object <- AddMetaData(object = object, metadata = meta.data)
+  # Set identity class and Project
+  if (!is.null(x = hdf5r::h5attributes(x = file)$project)) {
+    Project(object = object) <- hdf5r::h5attr(x = file, which = 'project')
+  }
+  if (file$exists(name = 'active.ident')) {
+    idents <- file[['active.ident']][]
+    names(x = idents) <- cells
+    Idents(object = object) <- idents
+  } else {
+    Idents(object = object) <- Project(object = object)
+  }
+  # Load misc
+  for (x in names(x = file[['misc']])) {
+    if (verbose) {
+      message("Adding miscellaneous data ", x)
+    }
+    Misc(object = object, slot = x) <- as.list(x = file[['misc']][[x]])
+  }
+  # Load tools
+  slot(object = object, name = 'tools') <- as.list(x = file[['tools']])
+  return(object)
+}
+
+#' Print index information for an h5Seurat Index (h5SI)
+#'
+#' @inheritParams base::print
+#'
+#' @return Invisibly returns \code{x}
+#'
+#' @importFrom cli symbol
+#' @importFrom crayon red green col_align
+#'
+#' @aliases print
+#'
+#' @rdname print.h5SI
+#' @method print h5SI
+#' @export
+#'
+#' @keywords internal
+#'
+print.h5SI <- function(x, ...) {
+  catn <- function(...) {
+    cat(..., '\n', sep = '')
+  }
+  symbols <- c(red(symbol$cross), green(symbol$tick))
+  x <- ValidateH5SI(x = x)
+  # Show assay information
+  catn("Assays:")
+  assays <- names(x = x$assays)
+  assays <- paste0(
+    ' ',
+    ifelse(test = assays == x$active.assay, yes = '*', no = ''),
+    assays,
+    ': '
+  )
+  assays <- col_align(
+    text = assays,
+    width = max(nchar(x = assays)),
+    align = 'right'
+  )
+  catn(
+    MakeSpace(n = max(nchar(x = assays))),
+    col_align(
+      text = c('counts', 'data', 'scale.data'),
+      width = nchar(x = 'scale.data') + 1,
+      align = 'center'
+    )
+  )
+  for (i in 1:length(x = x$assays)) {
+    catn(
+      assays[i],
+      col_align(
+        text = symbols[x$assays[[i]] + 1],
+        width = nchar(x = 'scale.data') + 1,
+        align = 'center'
+      )
+    )
+  }
+  # Show reduction informaiton
+  reduc.header <- c(
+    'Embeddings' = 'cell.embeddings',
+    'Loadings' = 'feature.loadings',
+    'Projected' = 'feature.loadings.projected',
+    'JackStraw' = 'jackstraw'
+  )
+  grouped.reductions <- split(
+    x = x$reductions,
+    f = unique(x = sapply(X = x$reductions, FUN = '[[', 'assay'))
+  )
+  for (assay in names(x = grouped.reductions)) {
+    catn("Reductions for ", assay, ":")
+    reduc.use <- grouped.reductions[[assay]]
+    reductions <- paste0(' ', names(x = reduc.use), ': ')
+    reductions <- col_align(text = reductions, width = max(nchar(x = reductions)))
+    catn(
+      MakeSpace(n = max(nchar(x = reductions))),
+      col_align(
+        text = names(x = reduc.header),
+        width = max(nchar(x = names(x = reduc.header))) + 1,
+        align = 'center'
+      )
+    )
+    for (i in 1:length(x = grouped.reductions)) {
+      catn(
+        reductions[i],
+        col_align(
+          text = symbols[unlist(x = reduc.use[[i]][reduc.header]) + 1],
+          width = max(nchar(x = names(x = reduc.header))) + 1,
+          align = 'center'
+        )
+      )
+    }
+  }
+  # Show graph information
+  for (assay in names(x = x$graphs)) {
+    catn("Graphs for ", assay, ":")
+    catn(' ', paste(x$graphs[[assay]], collapse = ', '))
+  }
+  return(invisible(x = x))
+}
+
+#' @importFrom Seurat as.Seurat Project
+#'
+#' @rdname SaveH5Seurat
+#' @method SaveH5Seurat default
+#' @export
+#'
+#' @keywords internal
+#'
+SaveH5Seurat.default <- function(
+  object,
+  filename,
+  overwrite = FALSE,
+  verbose = TRUE,
+  ...
+) {
+  tryCatch(
+    expr = object <- as.Seurat(x = object, verbose = verbose, ...),
+    error = function(...) {
+      stop(
+        "Unable to coerce an object of class ",
+        class(x = object),
+        " to a Seurat object",
+        call. = FALSE
+      )
+    }
+  )
+  if (missing(x = filename)) {
+    filename <- paste0(Project(object = object), '.h5Seurat')
+  }
+  return(invisible(x = SaveH5Seurat(
+    object = object,
+    filename = filename,
+    overwrite = overwrite,
+    verbose = verbose
+  )))
+}
+
+#' @importFrom tools file_ext
+#' @importFrom methods slot slotNames
+#' @importFrom Seurat Project DefaultAssay Idents GetAssayData Key
+#' VariableFeatures Misc Embeddings Loadings Stdev JS Command Tool
+#'
+#' @rdname SaveH5Seurat
+#' @method SaveH5Seurat Seurat
+#' @export
+#'
+#' @keywords internal
+#'
+SaveH5Seurat.Seurat <- function(
+  object,
+  filename = paste0(Project(object = object), '.h5Seurat'),
+  overwrite = FALSE,
+  verbose = TRUE,
+  ...
+) {
+  if (file_ext(x = filename) != 'h5Seurat') {
+    filename <- paste0(filename, '.h5Seurat')
+  }
+  if (file.exists(filename)) {
+    if (overwrite) {
+      warning(
+        "Overwriting previous file ",
+        filename,
+        call. = FALSE,
+        immediate. = TRUE
+      )
+      file.remove(filename)
+    } else {
+      stop("H5Seurat file at ", filename, " exists", call. = FALSE)
+    }
+  }
+  hfile <- hdf5r::H5File$new(filename = filename, mode = 'w-')
+  on.exit(expr = hfile$close_all())
+  # Add attributes for project, version, and default assay
+  hdf5r::h5attr(x = hfile, which = 'project') <- Project(object = object)
+  hdf5r::h5attr(x = hfile, which = 'version') <- as.character(x = slot(
+    object = object,
+    name = 'version'
+  ))
+  hdf5r::h5attr(x = hfile, which = 'active.assay') <- DefaultAssay(object = object)
+  # Add meta.data, identity class, and cell names
+  hfile[['meta.data']] <- object[[]]
+  hfile[['active.ident']] <- Idents(object = object)
+  hfile[['cell.names']] <- colnames(x = object)
+  # Prepare group information
+  required.groups <- c('assays', 'reductions', 'graphs', 'commands', 'tools', 'misc')
+  for (group in required.groups) {
+    hfile$create_group(name = group)
+  }
+  # Add Assays
+  assays <- Filter(
+    f = function(x) {
+      return(inherits(x = object[[x]], what = 'Assay'))
+    },
+    x = names(x = object)
+  )
+  assay.group <- hfile[['assays']]
+  for (assay in assays) {
+    obj <- object[[assay]]
+    obj.group <- assay.group$create_group(name = assay)
+    # Add counts/data/scale.data
+    for (i in c('counts', 'data', 'scale.data')) {
+      dat <- GetAssayData(object = obj, slot = i)
+      if (!IsMatrixEmpty(x = dat)) {
+        if (verbose) {
+          message("Adding ", i, " for ", assay)
+        }
+        if (inherits(x = dat, what = 'matrix')) {
+          obj.group[[i]] <- dat
+        } else if (inherits(x = dat, what = 'dgCMatrix')) {
+          dat.grp <- obj.group$create_group(name = i)
+          dat.grp[['indices']] <- slot(object = dat, name = 'i')
+          dat.grp[['indptr']] <- slot(object = dat, name = 'p')
+          dat.grp[['data']] <- slot(object = dat, name = 'x')
+        } else {
+          warning("wrong data")
+        }
+        if (i == 'scale.data') {
+          obj.group[['scaled.features']] <- rownames(x = dat)
+        }
+      } else if (verbose) {
+        message("No data found in ", i, " for ", assay)
+      }
+    }
+    # Add key
+    hdf5r::h5attr(x = obj.group, which = 'key') <- Key(object = obj)
+    # Add variable features
+    if (length(x = VariableFeatures(object = obj)) > 0) {
+      if (verbose) {
+        message("Adding variable features for ", assay)
+      }
+      obj.group[['variable.features']] <- VariableFeatures(object = obj)
+    } else if (verbose) {
+      message("No variable features found for ", assay)
+    }
+    # Add meta.features
+    if (ncol(x = obj[[]]) > 0) {
+      if (verbose) {
+        message("Adding meta.feature information for ", assay)
+      }
+      obj.group[['meta.features']] <- obj[[]]
+    } else if (verbose) {
+      message("No meta.feature information for ", assay)
+    }
+    # Add misc
+    if (length(x = Misc(object = obj)) > 0) {
+      misc.group <- obj.group$create_group(name = 'misc')
+      for (i in names(x = Misc(object = obj))) {
+        if (verbose) {
+          message("Adding miscellaneous data ", i, " for ", assay)
+        }
+        WriteH5List(
+          x = Misc(object = obj, slot = i),
+          name = i,
+          hgroup = misc.group
+        )
+        # misc.group[[i]] <- Misc(object = obj, slot = i)
+      }
+    } else if (verbose) {
+      message("No miscellaneous data found for ", assay)
+    }
+    # Add feature names
+    obj.group[['feature.names']] <- rownames(x = obj)
+  }
+  # Add DimReducs
+  reduc.group <- hfile[['reductions']]
+  reductions <- Filter(
+    f = function(x) {
+      return(inherits(x = object[[x]], what = 'DimReduc'))
+    },
+    x = names(x = object)
+  )
+  if (length(x = reductions) > 0) {
+    for (reduc in reductions) {
+      obj <- object[[reduc]]
+      obj.group <- reduc.group$create_group(name = reduc)
+      # Add cell embeddings
+      if (!IsMatrixEmpty(x = Embeddings(object = obj))) {
+        if (verbose) {
+          message("Adding cell embeddings for ", reduc)
+        }
+        obj.group[['cell.embeddings']] <- Embeddings(object = obj)
+      } else if (verbose) {
+        message("No cell embeddings for ", reduc)
+      }
+      # Add feature loadings
+      for (i in c('feature.loadings', 'feature.loadings.projected')) {
+        projected <- grepl(pattern = 'projected', x = i)
+        type <- ifelse(test = projected, yes = 'projected loadings', no = 'loadings')
+        if (!IsMatrixEmpty(x = Loadings(object = obj, projected = projected))) {
+          if (verbose) {
+            message("Adding ", type, " for ", reduc)
+          }
+          loadings <- Loadings(object = obj, projected = projected)
+          obj.group[[i]] <- loadings
+          features <- ifelse(
+            test = projected,
+            yes = 'projected.features',
+            no = 'features'
+          )
+          obj.group[[features]] <- rownames(x = loadings)
+        } else if (verbose) {
+          message("No ", type, " for ", reduc)
+        }
+      }
+      # Add default assay and key
+      hdf5r::h5attr(x = obj.group, which = 'active.assay') <- DefaultAssay(object = obj)
+      hdf5r::h5attr(x = obj.group, which = 'key') <- Key(object = obj)
+      # Add standard deviations
+      if (length(x = Stdev(object = obj)) > 0) {
+        if (verbose) {
+          message("Adding standard deviations for ", reduc)
+        }
+        obj.group[['stdev']] <- Stdev(object = obj)
+      } else if (verbose) {
+        message("No standard deviations for ", reduc)
+      }
+      # Add JackStraw
+      if (as.logical(x = JS(object = obj))) {
+        if (verbose) {
+          message("Adding JackStraw information for ", reduc)
+        }
+        js.group <- obj.group$create_group(name = 'jackstraw')
+        for (x in slotNames(x = JS(object = obj))) {
+          js.group[[x]] <- JS(object = obj, slot = x)
+        }
+      } else if (verbose) {
+        message("No JackStraw data for ", reduc)
+      }
+      # Add misc
+      # TODO: Add Misc method for DimReducs
+      misc.group <- obj.group$create_group(name = 'misc')
+      for (i in Enumerate(x = slot(object = obj, name = 'misc'))) {
+        if (verbose) {
+          message("Adding miscellaneous data: ", i$name)
+        }
+        misc.group[[i$name]] <- i$value
+      }
+    }
+  } else if (verbose) {
+    message("No dimensional reduction information present")
+  }
+  # Add Graphs
+  graph.group <- hfile[['graphs']]
+  graphs <- Filter(
+    f = function(x) {
+      return(inherits(x = object[[x]], what = 'Graph'))
+    },
+    x = names(x = object)
+  )
+  if (length(x = graphs) > 0) {
+    for (graph in graphs) {
+      if (verbose) {
+        message("Saving graph ", graph)
+      }
+      obj.group <- graph.group$create_group(name = graph)
+      obj.group[['indices']] <- slot(object = object[[graph]], name = 'i')
+      obj.group[['indptr']] <- slot(object = object[[graph]], name = 'p')
+      obj.group[['data']] <- slot(object = object[[graph]], name = 'x')
+    }
+  } else if (verbose) {
+    message("No nearest-neighbor graphs present")
+  }
+  # Add SeuratCommands
+  cmd.group <- hfile[['commands']]
+  for (i in Command(object = object)) {
+    if (verbose) {
+      message("Adding command ", i)
+    }
+    WriteH5List(
+      x = Filter(
+        f = Negate(f = is.function),
+        x = as.list(x = Command(object = object, command = i))
+      ),
+      name = i,
+      hgroup = cmd.group
+    )
+    cmd <- Command(object = object, command = i)
+    for (slot in slotNames(x = cmd)) {
+      if (slot == 'params') {
+        next
+      }
+      hdf5r::h5attr(x = cmd.group[[i]], which = slot) <- as.character(x = slot(
+        object = cmd,
+        name = slot
+      ))
+    }
+  }
+  # Add misc
+  misc.group <- hfile[['misc']]
+  for (m in Enumerate(x = Misc(object = object))) {
+    if (verbose) {
+      message("Adding miscellaneous data: ", m$name)
+    }
+    WriteH5List(x = m$value, name = m$name, hgroup = misc.group)
+  }
+  # Add tools
+  tool.group <- hfile[['tools']]
+  for (x in Tool(object = object)) {
+    if (verbose) {
+      message("Writing data for tool ", x)
+    }
+    WriteH5List(x = Tool(object = object, slot = x), name = x, hgroup = tool.group)
+  }
+  return(invisible(x = filename))
+}
+
+#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+# Standard functions
+#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+#' Assemble an object from an h5Seurat file
+#'
+#' @param assay,reduction,cmd Name of assay, reduction, or command to load
+#' @param file A connected h5Seurat file to pull the assay from
+#' @param cells A vector of cell names
+#' @param slots A vector of slots to load for the assay
+#' @param verbose Show progress updates
+#'
+#' @return \code{AssembleAssay}: An \code{Assay} object
+#'
+#' @rdname AssembleObject
+#' @name AssembleObject
+#'
+#' @importFrom methods slot<-
+#' @importFrom Seurat as.sparse CreateAssayObject Key<- GetAssayData
+#' SetAssayData AddMetaData VariableFeatures<-
+#'
+#' @keywords internal
+#'
+AssembleAssay <- function(
+  assay,
+  file,
+  cells,
+  slots = c('counts', 'data', 'scale.data'),
+  verbose = TRUE
+) {
+  assay.group <- file[['assays']][[assay]]
+  features <- assay.group[['feature.names']][]
+  if (!all(slots %in% c('counts', 'data', 'scale.data'))) {
+    stop("'slots' must be any or all of 'counts', 'data', or 'scale.data'", call. = FALSE)
+  } else if (!any(c('counts', 'data') %in% slots)) {
+    stop("At least one of 'counts' or 'data' must be in 'slots'", call. = FALSE)
+  }
+  # Add counts if not data, otherwise add data
+  if ('counts' %in% slots && !'data' %in% slots) {
+    if (verbose) {
+      message("Initializing ", assays, " with counts")
+    }
+    counts.data <- assay.group[['counts']]
+    counts <- if (inherits(x = counts.data, what = 'H5Group')) {
+      as.sparse(x = counts.data)
+    } else {
+      counts.data[, ]
+    }
+    rownames(x = counts) <- features
+    colnames(x = counts) <- cells
+    obj <- CreateAssayObject(
+      counts = counts,
+      min.cells = -1,
+      min.features = -1
+    )
+  } else {
+    if (verbose) {
+      message("Initializing ", assay, " with data")
+    }
+    data.data <- assay.group[['data']]
+    data <- if (inherits(x = data.data, what = 'H5Group')) {
+      as.sparse(x = data.data)
+    } else {
+      data.data[, ]
+    }
+    rownames(x = data) <- features
+    colnames(x = data) <- cells
+    obj <- CreateAssayObject(data = data)
+  }
+  Key(object = obj) <- hdf5r::h5attr(x = assay.group, which = 'key')
+  # Add remaining slots
+  for (slot in slots) {
+    if (IsMatrixEmpty(x = GetAssayData(object = obj, slot = slot))) {
+      if (verbose) {
+        message("Adding ", slot, " for ", assay)
+      }
+      dat.data <- assay.group[[slot]]
+      dat <- if (inherits(x = dat.data, what = 'H5Group')) {
+        as.sparse(x = dat.data)
+      } else {
+        dat.data[, ]
+      }
+      rownames(x = dat) <- if (slot == 'scale.data') {
+        assay.group[['scaled.features']][]
+      } else {
+        features
+      }
+      # colnames(x = dat) <- cells
+      colnames(x = dat) <- cells
+      obj <- SetAssayData(object = obj, slot = slot, new.data = dat)
+    }
+  }
+  # Add meta features
+  if (assay.group$exists(name = 'meta.features')) {
+    if (verbose) {
+      message("Adding feature-level metadata for ", assay)
+    }
+    meta.data <- assay.group[['meta.features']][]
+    rownames(x = meta.data) <- features
+    obj <- AddMetaData(
+      object = obj,
+      metadata = meta.data
+    )
+  } else if (verbose) {
+    message("No feature-level metadata for ", assay)
+  }
+  # Add variable feature information
+  if (assay.group$exists(name = 'variable.features')) {
+    if (verbose) {
+      message("Adding variable feature information for ", assay)
+    }
+    VariableFeatures(object = obj) <- assay.group[['variable.features']][]
+  } else if (verbose) {
+    message("No variable features for ", assay)
+  }
+  # Add miscellaneous information
+  if (assay.group$exists(name = 'misc')) {
+    slot(object = obj, name = 'misc') <- as.list(x = assay.group[['misc']])
+  } else if (verbose) {
+    message("No miscellaneous information for ", assay)
+  }
+  return(obj)
+}
+
+#' @param default.assay Name of default assay for this object
+#' @param embeddings,loadings,projected,jackstraw Pull cell embeddings, feature
+#' loadings, projected loadings, and/or JackStraw information
+#'
+#' @return \code{AssembleDimReduc}: A \code{DimReduc} object
+#'
+#' @rdname AssembleObject
+#'
+#' @importFrom Seurat CreateDimReducObject JS
+#'
+#' @keywords internal
+#'
+AssembleDimReduc <- function(
+  reduction,
+  file,
+  cells,
+  default.assay,
+  embeddings = TRUE,
+  loadings = FALSE,
+  projected = FALSE,
+  jackstraw = FALSE,
+  verbose = TRUE
+) {
+  reduc.group <- file[['reductions']][[reduction]]
+  key <- hdf5r::h5attr(x = reduc.group, which = 'key')
+  # Pull cell embeddings
+  if (embeddings) {
+    if (verbose) {
+      message("Pulling cell embeddings for ", reduction)
+    }
+    cell.embeddings <- reduc.group[['cell.embeddings']][, ]
+    rownames(x = cell.embeddings) <- cells
+    colnames(x = cell.embeddings) <- paste0(key, 1:ncol(x = cell.embeddings))
+  } else {
+    if (verbose) {
+      message("No cell embeddings for ", reduction)
+    }
+    cell.embeddings <- new(Class = 'matrix')
+  }
+  # Pull feature loadings
+  if (loadings) {
+    if (verbose) {
+      message("Pulling feature loadings for ", reduction)
+    }
+    feature.loadings <- reduc.group[['feature.loadings']][, ]
+    rownames(x = feature.loadings) <- reduc.group[['features']][]
+    colnames(x = feature.loadings) <- paste0(key, 1:ncol(x = feature.loadings))
+  } else {
+    if (verbose) {
+      message("No feature loadings for ", reduction)
+    }
+    feature.loadings <- new(Class = 'matrix')
+  }
+  # Pull projected loadings
+  if (projected) {
+    if (verbose) {
+      message("Pulling projected loadings for ", reduction)
+    }
+    projected.loadings <- reduc.group[['feature.loadings.projected']][, ]
+    rownames(x = projected.loadings) <- reduc.group[['projected.features']][]
+    colnames(x = projected.loadings) <- paste0(key, 1:ncol(x = projected.loadings))
+  } else {
+    if (verbose) {
+      message("No projected loadings for ", reduction)
+    }
+    projected.loadings <- new(Class = 'matrix')
+  }
+  obj <- CreateDimReducObject(
+    embeddings = cell.embeddings,
+    loadings = feature.loadings,
+    projected = projected.loadings,
+    # assay = index$reductions[[x]]$assay,
+    assay = default.assay,
+    stdev = if (reduc.group$exists(name = 'stdev')) {
+      reduc.group[['stdev']][]
+    } else {
+      numeric()
+    },
+    key = key
+  )
+  # Add misc
+  if (reduc.group$exists(name = 'misc')) {
+    if (verbose) {
+      message("Pulling miscellaneous information for ", reduction)
+    }
+    slot(object = obj, name = 'misc') <- as.list(x = reduc.group[['misc']])
+  } else if (verbose) {
+    message("No miscellaneous information for ", reduction)
+  }
+  # Add JackStraw
+  if (jackstraw) {
+    if (verbose) {
+      message("Pulling JackStraw information for ", reduction)
+    }
+    js <- new(Class = 'JackStrawData')
+    for (slot in names(x = reduc.group[['jackstraw']])) {
+      JS(object = js, slot = slot) <- as.matrix(x = reduc.group[['jackstraw']][[slot]][, ])
+    }
+    JS(object = obj) <- js
+  } else if (verbose) {
+    message("No JackStraw information for ", reduction)
+  }
+  return(obj)
+}
+
+#' @return \code{AssembleSeuratCommand}: A \code{SeuratCommand} object
+#'
+#' @rdname AssembleObject
+#'
+#' @keywords internal
+#'
+AssembleSeuratCommand <- function(cmd, file, verbose = TRUE) {
+  if (verbose) {
+    message("Loading command information for ", cmd)
+  }
+  return(new(
+    Class = 'SeuratCommand',
+    name = hdf5r::h5attr(x = file[['commands']][[cmd]], which = 'name'),
+    time.stamp = as.POSIXct(x = hdf5r::h5attr(
+      x = file[['commands']][[cmd]],
+      which = 'time.stamp'
+    )),
+    call.string = hdf5r::h5attr(
+      x = file[['commands']][[cmd]],
+      which = 'call.string'
+    ),
+    params = as.list(x = file[['commands']][[cmd]])
+  ))
+}
+
+#' Figure out which objects to load from an h5Seurat file
+#'
+#' @inheritParams LoadH5Seurat
+#' @param index An h5Seurat index (\code{h5SI}) from \code{\link{IndexH5Seurat}}
+#'
+#' @return \code{GetAssays}: A named list where each entry is a vector describing
+#' the slots of an assay to load and the names are the assays to load
+#'
+#' @seealso \code{\link{LoadH5Seurat}} \code{\link{IndexH5Seurat}}
+#'
+#' @rdname GetObject
+#' @name GetObject
+#'
+#' @keywords internal
+#'
+GetAssays <- function(assays, index) {
+  assay.slots <- c('counts', 'data', 'scale.data')
+  assay.msg <- 'Assay specification must include either the name of an assay or one or more assay slots'
+  if (!is.null(x = names(x = assays))) {
+    assays <- as.list(x = assays)
+  }
+  if (!is.list(x = assays)) {
+    if (any(assays %in% names(x = index$assays)) && any(assays %in% assay.slots)) {
+      stop("Ambiguous assays", call. = FALSE)
+    } else if (any(assays %in% names(x = index$assays))) {
+      assays <- assays[assays %in% names(x = index$assays)]
+      assays <- sapply(
+        X = assays,
+        FUN = function(...) {
+          return(assay.slots)
+        },
+        simplify = FALSE,
+        USE.NAMES = TRUE
+      )
+    } else {
+      assays <- assays[assays %in% assay.slots]
+      if (length(x = assays) < 1) {
+        stop(assay.msg, call. = FALSE)
+      }
+      assays <- list(assays)
+      assays <- rep_len(x = assays, length.out = length(x = index$assays))
+      names(x = assays) <- names(x = index$assays)
+    }
+  } else {
+    for (i in 1:length(x = assays)) {
+      assay.name <- names(x = assays)[i] %||% names(x = index$assays)[i] %||% ''
+      if (!assay.name %in% names(x = index$assays)) {
+        if (assays[[i]][1] %in% names(x = index$assays)) {
+          assay.name <- assays[[i]][1]
+        } else if (any(assays[[i]] %in% assay.slots)) {
+          assay.name <- hdf5r::h5attr(x = file, which = 'active.assay')
+        }
+      }
+      if (nchar(x = assay.name) < 0 || !assay.name %in% names(x = index$assays)) {
+        stop(assay.msg, call. = FALSE)
+      }
+      assay.content <- assays[[i]]
+      if (assay.content[1] %in% names(x = index$assays)) {
+        assay.content <- assay.slots
+      } else {
+        assay.content <- assay.content[assay.content %in% assay.slots]
+        if (length(x = assay.content) < 1) {
+          stop(assay.msg, call. = FALSE)
+        }
+      }
+      assays[i] <- list(assay.content)
+      names(x = assays)[i] <- assay.name
+    }
+  }
+  assays.checked <- assays
+  unique.assays <- unique(x = names(x = assays.checked))
+  assays <- vector(mode = 'list', length = length(x = unique.assays))
+  names(x = assays) <- unique.assays
+  for (i in unique.assays) {
+    assays.use <- which(x = names(x = assays.checked) == i)
+    slots.use <- unique(x = unlist(x = assays.checked[assays.use], use.names = FALSE))
+    slots.use <- slots.use[match(x = names(x = index$assays[[i]]), table = slots.use)]
+    slots.use <- as.character(x = na.omit(object = slots.use[index$assays[[i]]]))
+    assays[[i]] <- slots.use
+  }
+  for (i in Enumerate(x = assays)) {
+    if (!any(c('counts', 'data') %in% i$value)) {
+      stop("All assays must have either a counts or data slot, missing for ", i$name, call. = FALSE)
+    }
+  }
+  return(assays)
+}
+
+#' @param reductions A vector of reduction names to load
+#'
+#' @return \code{GetDimReducs}: A vector of reduction names that are derived from
+#' an assay in \code{assays}
+#'
+#' @rdname GetObject
+#'
+#' @keywords internal
+#'
+GetDimReducs <- function(reductions, index, assays = NULL) {
+  assays <- assays %||% names(x = index$assays)
+  assays <- GetAssays(assays = assays, index = index)
+  return(Filter(
+    f = function(reduc) {
+      return(index$reductions[[reduc]]$assay %in% names(x = assays))
+    },
+    x = reductions
+  ))
+}
+
+#' @param graphs A vector of graph names to load
+#'
+#' @return \code{GetGraphs}: A vector of graph names that are derived from an
+#' assay in \code{assays}
+#'
+#' @rdname GetObject
+#'
+#' @keywords internal
+#'
+GetGraphs <- function(index, assays = NULL) {
+  assays <- assays %||% names(x = index$assays)
+  assays <- GetAssays(assays = assays, index = index)
+  graph.assays <- vapply(
+    X = names(x = index$graphs),
+    FUN = function(x) {
+      return(any(grepl(pattern = x, x = names(x = assays), ignore.case = TRUE)))
+    },
+    FUN.VALUE = logical(length = 1L)
+  )
+  return(unlist(x = index$graphs[graph.assays]))
+}
+
+#' Construct an h5Seurat index
+#'
+#' @param ... ...
+#'
+#' @return ...
+#'
+#' @keywords internal
+#'
+h5SI <- function(...) {
+  object <- list(
+    'assays' = list(),
+    'reductions' = list(),
+    'graphs' = list(),
+    'active.assay' = vector(mode = 'character', length = 1L),
+    'meta.data' = vector(mode = 'character'),
+    ''
+  )
+  return(structure(.Data = object, class = 'h5SI'))
+}
+
+#' Validate an h5Seurat Index
+#'
+#' @param x An h5Seurat Index (h5SI)
+#'
+#' @return \code{x} if valid, otherwise a modified \code{x} to be valid
+#'
+#' @keywords internal
+#'
+ValidateH5SI <- function(x) {
+  if (!inherits(x = x, what = 'h5SI')) {
+    stop("'x' must be an h5Seurat index", call. = FALSE)
+  }
+  return(x)
+}
+
+#' Write a list as a series of HDF5 groups and datasets
+#'
+#' @param x A list
+#' @param name Name to save dataset as
+#' @param hgroup An HDF5 file or group (\code{H5File} or
+#' \code{H5Group} objects from hdf5r)
+#'
+#' @return Invisibly returns \code{NULL}
+#'
+#' @importFrom methods slotNames slot
+#'
+#' @keywords internal
+#'
+WriteH5List <- function(x, name, hgroup) {
+  if (!inherits(x = hgroup, what = c('H5File', 'H5Group'))) {
+    stop("'hgroup' must be an H5File or H5Group object from hdf5r", call. = FALSE)
+  }
+  if (is.list(x = x) && !is.data.frame(x = x)) {
+    xgroup <- hgroup$create_group(name = name)
+    for (i in Enumerate(x = x)) {
+      if ((is.list(x = i$value) && !is.data.frame(x = i$value)) || isS4(i$value)) {
+        WriteH5List(x = i$value, name = i$name, hgroup = xgroup)
+      } else if (!is.null(x = i$value)) {
+        # message(i$name, ': ', class(i$value))
+        xgroup[[i$name]] <- i$value
+      }
+    }
+  } else if (isS4(x)) {
+    xgroup <- hgroup$create_group(name = name)
+    hdf5r::h5attr(x = xgroup, which = 's4class') <- class(x = x)[1]
+    for (i in slotNames(x = x)) {
+      obj <- slot(object = x, name = i)
+      if (is.list(x = obj) && !is.data.frame(x = obj)) {
+        WriteH5List(x = obj, name = i, hgroup = xgroup)
+      } else if (!is.null(x = obj)) {
+        xgroup[[i]] <- obj
+      }
+    }
+  } else if (!is.null(x = x)) {
+    hgroup[[name]] <- x
+  }
+  return(invisible(x = NULL))
+}
diff --git a/R/seurat_data.R b/R/seurat_data.R
index ff1743d..f27ca04 100644
--- a/R/seurat_data.R
+++ b/R/seurat_data.R
@@ -12,11 +12,13 @@ NULL
 #'   \item{InstalledVersion}{Version of dataset installed, \code{NA} if not installed}
 #' }
 #'
-#' Other columns are extra metadata, and may change as new datasets are made available
-#'
 #' @export
 #'
-#' @seealso \code{\link{InstallData}} \code{\link{InstalledData}} \code{\link{RemoveData}} \code{\link{UpdateData}}
+#' @seealso \code{\link{InstallData}} \code{\link{InstalledData}}
+#' \code{\link{RemoveData}} \code{\link{UpdateData}}
+#'
+#' @examples
+#' AvailableData()
 #'
 AvailableData <- function() {
   UpdateManifest()
@@ -35,10 +37,22 @@ AvailableData <- function() {
 #'
 #' @export
 #'
-#' @seealso \code{\link{AvailableData}} \code{\link{InstalledData}} \code{\link{RemoveData}} \code{\link{UpdateData}}
+#' @seealso \code{\link{AvailableData}} \code{\link{InstalledData}}
+#' \code{\link{RemoveData}} \code{\link{UpdateData}}
+#'
+#' @examples
+#' \dontrun{
+#' InstallData('pbmc3k')
+#' }
 #'
 InstallData <- function(ds, force.reinstall = FALSE, ...) {
   UpdateManifest()
+  if (pkg.env$source != 'remote') {
+    stop(
+      "No access to remote SeuratData repository, unable to install new datasets",
+      call. = FALSE
+    )
+  }
   pkgs <- NameToPackage(ds = ds)
   if (!force.reinstall) {
     installed <- intersect(x = pkgs, y = rownames(x = InstalledData()))
@@ -46,7 +60,7 @@ InstallData <- function(ds, force.reinstall = FALSE, ...) {
       warning(
         "The following packages are already installed and will not be reinstalled: ",
         paste(
-          gsub(pattern = '\\.SeuratData', replacement = '', x = installed),
+          gsub(pattern = pkg.key, replacement = '', x = installed),
           collapse = ', '
         ),
         call. = FALSE,
@@ -63,7 +77,12 @@ InstallData <- function(ds, force.reinstall = FALSE, ...) {
   for (p in pkgs2[pkgs2 %in% search()]) {
     detach(name = p, unload = TRUE, character.only = TRUE)
   }
-  install.packages(pkgs = pkgs, repos = getOption(x = "SeuratData.repo.use"), type = 'source', ...)
+  install.packages(
+    pkgs = pkgs,
+    repos = getOption(x = "SeuratData.repo.use"),
+    type = 'source',
+    ...
+  )
   for (pkg in pkgs) {
     attachNamespace(ns = pkg)
     pkg.env$attached <- c(pkg.env$attached, pkg)
@@ -86,13 +105,114 @@ InstallData <- function(ds, force.reinstall = FALSE, ...) {
 #'
 #' @export
 #'
-#' @seealso \code{\link{AvailableData}} \code{\link{InstallData}} \code{\link{RemoveData}} \code{\link{UpdateData}}
+#' @seealso \code{\link{AvailableData}} \code{\link{InstallData}}
+#' \code{\link{RemoveData}} \code{\link{UpdateData}}
+#'
+#' @examples
+#' InstalledData()
 #'
 InstalledData <- function() {
   dat <- AvailableData()
   return(dat[which(x = dat$Installed, ), , drop = FALSE])
 }
 
+#' Modularly load a dataset
+#'
+#' @inheritParams AppendData
+#' @inheritParams LoadH5Seurat
+#' @param ds Name of dataset to load
+#' @param type How to load the \code{Seurat} object; choose from
+#' \describe{
+#'   \item{info}{
+#'   Information about the object and what's stored in it; to specify a dataset
+#'   to get information for, pass 'info_\code{name of dataset}' based on the name
+#'   of the dataset in the \code{other.datasets} column of the data manifest
+#'   }
+#'   \item{raw}{The raw form of the dataset, no other options are evaluated}
+#'   \item{...}{
+#'   Name of dataset in the \code{other.datasets} column of the data manifest
+#'   corresponding to the dataset desired (eg. 'processed' for ifnb)
+#'   }
+#' }
+#'
+#' @inherit LoadH5Seurat return
+#'
+#' @importFrom utils data
+#'
+#' @export
+#'
+#' @seealso \code{\link[utils]{data}}
+#'
+#' @examples
+#' \dontrun{
+#' LoadData('ifnb')
+#' ifnb.raw <- LoadData('ifnb', type = 'raw')
+#' ifnb.proccessed <- LoadData('ifnb', type = 'processed')
+#' }
+#'
+LoadData <- function(
+  ds,
+  type = 'info',
+  assays = NULL,
+  reductions = NULL,
+  graphs = NULL,
+  object = NULL,
+  overwrite = FALSE,
+  verbose = TRUE
+) {
+  ds <- ds[1]
+  installed <- InstalledData()
+  if (!NameToPackage(ds = ds) %in% rownames(x = installed)) {
+    stop("Cannot find dataset ", ds, call. = FALSE)
+  }
+  ds <- NameToPackage(ds = ds)
+  choices <- na.omit(object = installed[ds, 'other.datasets', drop = TRUE])
+  info <- grepl(pattern = '^info', x = type)
+  type <- gsub(pattern = '^info_', replacement = '', x = type)
+  type <- match.arg(arg = tolower(x = type), choices = c('info', 'raw', choices))
+  if (type == 'raw') {
+    e <- new.env()
+    ds <- gsub(pattern = '\\.SeuratData', replacement = '', x = ds)
+    data(list = ds, envir = e)
+    return(e[[ds]])
+  }
+  if (length(x = choices) < 1) {
+    stop(
+      "No alternate datasets available for ",
+      gsub(pattern = '.SeuratData$', replacement = '', x = ds),
+      call. = FALSE
+    )
+  }
+  type <- switch(
+    EXPR = type,
+    'info' = choices[1],
+    type
+  )
+  filename <- file.path('extdata', 'objects', paste0(type, '.h5Seurat'))
+  filename <- system.file(filename, package = ds, mustWork = TRUE)
+  object <- if (is.null(x = object) && !info) {
+    LoadH5Seurat(
+      file = filename,
+      type = ifelse(test = info, yes = 'info', no = 'object'),
+      assays = assays,
+      reductions = reductions,
+      graphs = graphs,
+      verbose = verbose
+    )
+  } else {
+    AppendData(
+      file = filename,
+      object = object,
+      assays = assays,
+      reductions = reductions,
+      graphs = graphs,
+      overwrite = overwrite,
+      verbose = verbose
+    )
+  }
+  return(object)
+}
+
 #' Remove a dataset
 #'
 #' @inheritParams utils::remove.packages
@@ -102,7 +222,13 @@ InstalledData <- function() {
 #'
 #' @export
 #'
-#' @seealso \code{\link{AvailableData}} \code{\link{InstallData}} \code{\link{InstalledData}} \code{\link{UpdateData}}
+#' @seealso \code{\link{AvailableData}} \code{\link{InstallData}}
+#' \code{\link{InstalledData}} \code{\link{UpdateData}}
+#'
+#' @examples
+#' \dontrun{
+#' RemoveData('pbmc3k')
+#' }
 #'
 RemoveData <- function(ds, lib) {
   UpdateManifest()
@@ -125,9 +251,22 @@ RemoveData <- function(ds, lib) {
 #'
 #' @export
 #'
-#' @seealso \code{\link{AvailableData}} \code{\link{InstallData}} \code{\link{InstalledData}} \code{\link{RemoveData}}
+#' @seealso \code{\link{AvailableData}} \code{\link{InstallData}}
+#' \code{\link{InstalledData}} \code{\link{RemoveData}}
+#'
+#' @examples
+#' \dontrun{
+#' UpdateData(ask = FALSE)
+#' }
 #'
 UpdateData <- function(ask = TRUE, lib.loc = NULL) {
+  UpdateManifest()
+  if (pkg.env$source != 'remote') {
+    stop(
+      "No access to remote SeuratData repository, unable to update datasets",
+      call. = FALSE
+    )
+  }
   update.packages(lib.loc = lib.loc, repos = getOption(x = "SeuratData.repo.use"), ask = ask, type = 'source')
   UpdateManifest()
   invisible(x = NULL)
diff --git a/R/zzz.R b/R/zzz.R
index 156cf16..fcc5638 100644
--- a/R/zzz.R
+++ b/R/zzz.R
@@ -2,11 +2,19 @@
 #'
 #' @section Package options:
 #'
-#' SeuratData uses the following [options()] to configure behaviour:
+#' SeuratData uses the following options to control behaviour, users can configure
+#' these with \code{\link[base]{options}}:
 #'
 #' \itemize{
 #'   \item `SeuratData.repo.use`: Set the location where the SeuratData datasets
 #'   are stored. Users generally should not modify.
+#'   \item `SeuratData.manifest.cache`: Cache the data manifest whenever we talk
+#'   to the data repository; note, setting to \code{FALSE} will simply prevent
+#'   SeuratData from caching the manifest, not from reading a previously cached
+#'   manifest
+#'   \item `SeuratData.roaming`: For Windows users, use a roaming profile directory
+#'   for domain users. See \url{https://en.wikipedia.org/wiki/Roaming_user_profile}
+#'   for a brief overview and Microsoft's documentation for greater detail
 #' }
 #'
 #' @docType package
@@ -17,16 +25,23 @@
 "_PACKAGE"
 
 #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-# Global variables and environment
+# Global variables, default options, and package environment
 #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 
 default.options <- list(
-  SeuratData.repo.use = 'https://seurat.nygenome.org'
+  SeuratData.repo.use = 'http://seurat.nygenome.org/',
+  SeuratData.manifest.cache = TRUE,
+  SeuratData.roaming = FALSE
 )
 
+pkg.key <- '\\.SeuratData$'
+
 pkg.env <- new.env()
-pkg.env$manifest <- vector(mode = 'character')
+pkg.env$manifest <- vector(mode = 'list')
+pkg.env$source <- vector(mode = 'character')
+pkg.env$update.call <- vector(mode = 'character')
 pkg.env$attached <- vector(mode = 'character')
+pkg.env$extdata.warn <- FALSE
 
 #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 # Internal functions
@@ -44,6 +59,12 @@ pkg.env$attached <- vector(mode = 'character')
 #' @author Hadley Wickham
 #' @references \url{https://adv-r.hadley.nz/functions.html#missing-arguments}
 #'
+#' @examples
+#' \dontrun{
+#' 4 %||% 5
+#' NULL %|| 5
+#' }
+#'
 #' @keywords internal
 #'
 `%||%` <- function(lhs, rhs) {
@@ -54,6 +75,28 @@ pkg.env$attached <- vector(mode = 'character')
   }
 }
 
+#' Get an application data directory
+#'
+#' @inheritParams AttachData
+#' @param author Author name for application
+#'
+#' @return A character vector with path to the application data dir
+#'
+#' @importFrom rappdirs user_data_dir
+#'
+#' @seealso \code{\link[rappdirs]{user_data_dir}}
+#'
+#' @keywords internal
+#'
+AppData <- function(pkgname = 'SeuratData', author = pkgname) {
+  return(user_data_dir(
+    appname = pkgname,
+    appauthor = author,
+    version = file.path('%p-platform', '%v'),
+    roaming = getOption(x = 'SeuratData.roaming', default = FALSE)
+  ))
+}
+
 #' Attach datasets
 #'
 #' @param pkgname Name of package
@@ -103,13 +146,7 @@ AttachData <- function(pkgname = 'SeuratData') {
     col_nchar(x = header) - max(col_nchar(x = pkgs[-col.1])),
     col_nchar(x = header) * (1 / 2)
   ))
-  space <- paste(
-    rep_len(
-      x = ' ',
-      length.out = max(space.start - max(col_nchar(x = pkgs[col.1])), 1)
-    ),
-    collapse = ''
-  )
+  space <- MakeSpace(n = max(space.start - max(col_nchar(x = pkgs[col.1])), 1))
   info <- paste0(pkgs[col.1], space, pkgs[-col.1])
   packageStartupMessage(paste(info, collapse = '\n'), '\n')
   packageStartupMessage(rule(center = 'Key'))
@@ -133,12 +170,99 @@ AttachData <- function(pkgname = 'SeuratData') {
   invisible(x = NULL)
 }
 
+#' Enumerate a list or vector
+#'
+#' @param x A list or a vector
+#'
+#' @return A list of length \code{x} with the following named values:
+#' \describe{
+#'   \item{\code{name}}{The name of \code{x} at a given index}
+#'   \item{\code{value}}{The value of \code{x} at the corresponding index}
+#' }
+#'
+#' @note For any given index \code{i} in \code{x}, all attempts to use the name
+#' of the value of \code{x} at \code{i} will be made. If there is no name
+#' (eg. \code{nchar(x = names(x = x)[i]) == 0}), the index \code{i} will be used
+#' in its stead
+#'
+#' @keywords internal
+#'
+Enumerate <- function(x) {
+  indices <- seq_along(along.with = x)
+  keys <- names(x = x) %||% as.character(x = indices)
+  keys[nchar(x = keys) == 0] <- indices[nchar(x = keys) == 0]
+  vals <- lapply(
+    X = indices,
+    FUN = function(i) {
+      return(c('name' = keys[i], 'value' = unname(obj = x[i])))
+    }
+  )
+  return(vals)
+}
+
+#' Check to see if a matrix is empty
+#'
+#' Deterime if a matrix is empty or not
+#'
+#' @param x A matrix
+#'
+#' @return \code{TRUE} if the matrix is empty otherwise \code{FALSE}
+#'
+#' @details A matrix is considered empty if it satisfies one of the following
+#' conditions:
+#' \itemize{
+#'   \item The dimensions of the matrix are 0-by-0 (\code{all(dim(x) == 0)})
+#'   \item The dimensions of the matrix are 1-by-1 (\code{all(dim(x) == 1)}) and
+#'   the sole value is a \code{NA}
+#' }
+#' These two situations correspond to matrices generated with either
+#' \code{new('matrix')} or \code{matrix()}
+#'
+#' @examples
+#' \dontrun{
+#' IsMatrixEmpty(new('matrix'))
+#' IsMatrixEmpty(matrix())
+#' IsMatrixEmpty(matrix(1:9, nrow = 3))
+#' }
+#'
+#' @keywords internal
+#'
+IsMatrixEmpty <- function(x) {
+  matrix.dims <- dim(x = x)
+  matrix.na <- all(matrix.dims == 1) && all(is.na(x = x))
+  return(all(matrix.dims == 0) || matrix.na)
+}
+
+#' Make a space
+#'
+#' @param n Length space should be
+#'
+#' @return A space (' ') of length \code{n}
+#'
+#' @examples
+#' \dontrun{
+#' MakeSpace(10)
+#' }
+#'
+#' @keywords internal
+#'
+MakeSpace <- function(n) {
+  return(paste(rep_len(x = ' ', length.out = n), collapse = ''))
+}
+
 #' Find dataset packages based on name
 #'
 #' @param ds Names of datasets
 #'
 #' @return A vector of package names based on dataset names
 #'
+#' @examples
+#' \dontrun{
+#' NameToPackage('cbmc')
+#' NameToPackage('pbmc3k.SeuratData')
+#' NameToPackage('notadataset')
+#' }
+#'
 #' @keywords internal
 #'
 NameToPackage <- function(ds) {
@@ -171,77 +295,134 @@ NameToPackage <- function(ds) {
 #' @keywords internal
 #'
 UpdateManifest <- function() {
-  avail.pkgs <- available.packages(
-    repos = getOption(x = "SeuratData.repo.use"),
-    type = 'source',
-    fields = c('Description', 'Title'),
-    ignore_repo_cache = TRUE
+  # Set some defaults
+  pkg.env$source <- 'remote'
+  cache.manifest <- file.path(
+    AppData(pkgname = 'SeuratData', author = 'Satija Lab'),
+    'manifest.Rds'
   )
-  avail.pkgs <- as.data.frame(x = avail.pkgs, stringsAsFactors = FALSE)
-  avail.pkgs <- avail.pkgs[grepl(pattern = '\\.SeuratData$', x = avail.pkgs$Package), , drop = FALSE]
-  avail.pkgs <- apply(
-    X = avail.pkgs,
-    MARGIN = 1,
-    FUN = function(pkg) {
-      dataset <- gsub(
-        pattern = '\\.SeuratData$',
-        replacement = '',
-        x = pkg[['Package']]
-      )
-      desc <- unlist(x = strsplit(x = pkg[['Description']], split = '\n'))
-      desc <- sapply(
-        X = strsplit(x = desc, split = ':'),
-        FUN = function(x) {
-          name <- trimws(x = x[[1]])
-          val <- trimws(x = unlist(x = strsplit(x = x[[2]], split = ',')))
-          val <- paste(val, collapse = ', ')
-          names(x = val) <- name
-          return(val)
-        }
-      )
-      desc <- sapply(
-        X = desc,
-        FUN = function(x) {
-          x <- tryCatch(
-            expr = as.numeric(x = x),
-            warning = function(...) {
-              return(x)
-            }
-          )
-          if (!is.numeric(x = x) && !is.na(x = as.logical(x = x))) {
-            x <- as.logical(x = x)
-          }
-          return(x)
-        },
-        simplify = FALSE,
-        USE.NAMES = TRUE
-      )
-      desc <- c(
-        'Dataset' = dataset,
-        'Version' = pkg[['Version']],
-        'Summary' = pkg[['Title']],
-        desc
+  # Attempt to get the manifest from the remote server
+  avail.pkgs <- tryCatch(
+    expr = available.packages(
+      repos = getOption(x = "SeuratData.repo.use"),
+      type = 'source',
+      fields = c('Description', 'Title'),
+      ignore_repo_cache = TRUE
+    ),
+    warning = function(...) {
+      pkg.env$source <- ifelse(
+        test = file.exists(cache.manifest),
+        yes = 'appdir',
+        no = 'extdata'
       )
-      return(desc)
     }
   )
-  manifest.names <- unique(x = unlist(
-    x = lapply(X = avail.pkgs, FUN = names),
-    use.names = FALSE
-  ))
-  for (pkg in names(x = avail.pkgs)) {
-    for (col in manifest.names) {
-      avail.pkgs[[pkg]][[col]] <- avail.pkgs[[pkg]][[col]] %||% NA
+  # Process the manifest
+  if (pkg.env$source == 'remote') {
+    # Lots of stuff to get the manifest modified from the available.packages format
+    # into something usable by SeuratData
+    pkg.env$extdata.warn <- FALSE
+    avail.pkgs <- as.data.frame(x = avail.pkgs, stringsAsFactors = FALSE)
+    # Ensure we only use datasets tagged with .SeuratData
+    avail.pkgs <- avail.pkgs[grepl(pattern = pkg.key, x = avail.pkgs$Package), , drop = FALSE]
+    # Filter down to dataset name, short summary from package title, and
+    # metadata contained in package description
+    avail.pkgs <- apply(
+      X = avail.pkgs,
+      MARGIN = 1,
+      FUN = function(pkg) {
+        # Get dataset name
+        dataset <- gsub(
+          pattern = pkg.key,
+          replacement = '',
+          x = pkg[['Package']]
+        )
+        # Process the description metadata
+        desc <- unlist(x = strsplit(x = pkg[['Description']], split = '\n'))
+        desc <- sapply(
+          X = strsplit(x = desc, split = ':'),
+          FUN = function(x) {
+            name <- trimws(x = x[[1]])
+            val <- trimws(x = unlist(x = strsplit(x = x[[2]], split = ',')))
+            val <- paste(val, collapse = ', ')
+            names(x = val) <- name
+            return(val)
+          }
+        )
+        desc <- sapply(
+          X = desc,
+          FUN = function(x) {
+            x <- tryCatch(
+              expr = as.numeric(x = x),
+              warning = function(...) {
+                return(x)
+              }
+            )
+            if (!is.numeric(x = x) && !is.na(x = as.logical(x = x))) {
+              x <- as.logical(x = x)
+            }
+            return(x)
+          },
+          simplify = FALSE,
+          USE.NAMES = TRUE
+        )
+        # Assemble the information
+        desc <- c(
+          'Dataset' = dataset,
+          'Version' = pkg[['Version']],
+          'Summary' = pkg[['Title']],
+          desc
+        )
+        return(desc)
+      }
+    )
+    # Pad missing metadata with NAs
+    manifest.names <- unique(x = unlist(
+      x = lapply(X = avail.pkgs, FUN = names),
+      use.names = FALSE
+    ))
+    for (pkg in names(x = avail.pkgs)) {
+      for (col in manifest.names) {
+        avail.pkgs[[pkg]][[col]] <- avail.pkgs[[pkg]][[col]] %||% NA
+      }
+    }
+    # Convert each entry to a dataframe and bind everything together
+    avail.pkgs <- lapply(X = avail.pkgs, FUN = as.data.frame, stringsAsFactors = FALSE)
+    avail.pkgs <- do.call(what = 'rbind', args = avail.pkgs)
+    # Coerce version information to package_version
+    avail.pkgs$Version <- package_version(x = avail.pkgs$Version)
+  } else if (pkg.env$source == 'appdir') {
+    # Read cached manifest
+    pkg.env$extdata.warn <- FALSE
+    packageStartupMessage(
+      "Using cached data manifest, last updated at ",
+      file.info(cache.manifest)$ctime
+    )
+    avail.pkgs <- readRDS(file = cache.manifest)
+  } else if (pkg.env$source == 'extdata') {
+    # Read SeuratData-bundled manifest
+    if (!pkg.env$extdata.warn) {
+      warning(
+        "Using SeuratData-bundled data manifest. ",
+        "This may be out-of-date and not contain the latest datasets. ",
+        "This warning will be shown once per session or until we can read from a remote or cached data manifest",
+        call. = FALSE,
+        immediate. = TRUE
+      )
     }
+    pkg.env$extdata.warn <- TRUE
+    avail.pkgs <- readRDS(file = system.file(
+      'extdata/manifest.Rds',
+      package = 'SeuratData',
+      mustWork = TRUE
+    ))
   }
-  avail.pkgs <- lapply(X = avail.pkgs, FUN = as.data.frame, stringsAsFactors = FALSE)
-  avail.pkgs <- do.call(what = 'rbind', args = avail.pkgs)
-  avail.pkgs$Version <- package_version(x = avail.pkgs$Version)
+  # Get dataset installation status
   avail.pkgs$Installed <- vapply(
     X = rownames(x = avail.pkgs),
     FUN = requireNamespace,
     FUN.VALUE = logical(length = 1L),
-    quietly = TRUE,
+    quietly = TRUE
   )
   avail.pkgs$InstalledVersion <- sapply(
     X = rownames(x = avail.pkgs),
@@ -255,14 +436,30 @@ UpdateManifest <- function() {
       return(as.character(x = pkg.version))
     }
   )
+  # Coerce version information to package_version
+  # Allow NAs to become effectively NA_pacakge_version_
   avail.pkgs$InstalledVersion <- package_version(
     x = avail.pkgs$InstalledVersion,
     strict = FALSE
   )
-  # TODO: remove these when we allow loading of processed datasets
-  ds.index <- which(x = colnames(x = avail.pkgs) %in% c('default.dataset', 'other.datasets'))
-  avail.pkgs <- avail.pkgs[, -ds.index]
+  # # TODO: remove these when we allow loading of processed datasets
+  # ds.index <- which(x = colnames(x = avail.pkgs) %in% c('default.dataset', 'other.datasets'))
+  # avail.pkgs <- avail.pkgs[, -ds.index]
   pkg.env$manifest <- avail.pkgs
+  # Cache the manifest
+  if (getOption(x = 'SeuratData.manifest.cache', default = FALSE)) {
+    if (!dir.exists(paths = dirname(path = cache.manifest))) {
+      dir.create(path = dirname(path = cache.manifest), recursive = TRUE)
+    }
+    cached <- if (file.exists(cache.manifest)) {
+      readRDS(file = cache.manifest)
+    } else {
+      NULL
+    }
+    if (!isTRUE(x = all.equal(target = pkg.env$manifest, current = cached))) {
+      saveRDS(object = pkg.env$manifest, file = cache.manifest)
+    }
+  }
   invisible(x = NULL)
 }
 
diff --git a/inst/extdata/manifest.Rds b/inst/extdata/manifest.Rds
new file mode 100644
index 0000000..5dda2b0
Binary files /dev/null and b/inst/extdata/manifest.Rds differ
diff --git a/man/AppData.Rd b/man/AppData.Rd
new file mode 100644
index 0000000..eefc803
--- /dev/null
+++ b/man/AppData.Rd
@@ -0,0 +1,23 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/zzz.R
+\name{AppData}
+\alias{AppData}
+\title{Get an application data directory}
+\usage{
+AppData(pkgname = "SeuratData", author = pkgname)
+}
+\arguments{
+\item{pkgname}{Name of package}
+
+\item{author}{Author name for application}
+}
+\value{
+A character vector with path to the application data dir
+}
+\description{
+Get an application data directory
+}
+\seealso{
+\code{\link[rappdirs]{user_data_dir}}
+}
+\keyword{internal}
diff --git a/man/AppendData.Rd b/man/AppendData.Rd
new file mode 100644
index 0000000..9be5f9f
--- /dev/null
+++ b/man/AppendData.Rd
@@ -0,0 +1,62 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{AppendData}
+\alias{AppendData}
+\alias{AppendData.character}
+\alias{AppendData.H5File}
+\title{Append data from an h5Seurat file to a preexisting \code{Seurat} object}
+\usage{
+AppendData(file, object, ...)
+
+\method{AppendData}{character}(file, object, assays = NULL,
+  reductions = NULL, graphs = NULL, overwrite = FALSE,
+  verbose = TRUE, ...)
+
+\method{AppendData}{H5File}(file, object, assays = NULL,
+  reductions = NULL, graphs = NULL, overwrite = FALSE,
+  verbose = TRUE, ...)
+}
+\arguments{
+\item{file}{Name of h5Seurat or connected h5Seurat file to load}
+
+\item{object}{A \code{Seurat} object to append data to}
+
+\item{...}{Arguments passed to other methods}
+
+\item{assays}{One of:
+\itemize{
+  \item A character vector with names of assays
+  \item A character vector with one or more of \code{counts}, \code{data},
+  \code{scale.data} describing which slots of \strong{all assays} to load
+  \item A named list where each entry is either the name of an assay or a vector
+  describing which slots (described above) to take from which assay
+  \item \code{NULL} for all assays
+}}
+
+\item{reductions}{One of:
+\itemize{
+  \item A character vector with names of reductions
+  \item \code{NULL} for all reductions
+  \item \code{NA} or \code{FALSE} for no reductions
+}
+\strong{Note}: Only reductions associated with an assay loaded in \code{assays}}
+
+\item{graphs}{One of:
+\itemize{
+  \item A character vector with names of graphs
+  \item \code{NULL} for all graphs
+  \item \code{NA} or \code{FALSE} for no reductions
+}}
+
+\item{verbose}{Show progress updates}
+}
+\value{
+A \code{Seurat} object with the extra data requested
+}
+\description{
+Append data from an h5Seurat file to a preexisting \code{Seurat} object
+}
+\seealso{
+\code{\link{LoadH5Seurat}} \code{\link{LoadData}}
+}
+\keyword{internal}
diff --git a/man/AssembleObject.Rd b/man/AssembleObject.Rd
new file mode 100644
index 0000000..4bb316b
--- /dev/null
+++ b/man/AssembleObject.Rd
@@ -0,0 +1,45 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{AssembleObject}
+\alias{AssembleObject}
+\alias{AssembleAssay}
+\alias{AssembleDimReduc}
+\alias{AssembleSeuratCommand}
+\title{Assemble an object from an h5Seurat file}
+\usage{
+AssembleAssay(assay, file, cells, slots = c("counts", "data",
+  "scale.data"), verbose = TRUE)
+
+AssembleDimReduc(reduction, file, cells, default.assay,
+  embeddings = TRUE, loadings = FALSE, projected = FALSE,
+  jackstraw = FALSE, verbose = TRUE)
+
+AssembleSeuratCommand(cmd, file, verbose = TRUE)
+}
+\arguments{
+\item{assay, reduction, cmd}{Name of assay, reduction, or command to load}
+
+\item{file}{A connected h5Seurat file to pull the assay from}
+
+\item{cells}{A vector of cell names}
+
+\item{slots}{A vector of slots to load for the assay}
+
+\item{verbose}{Show progress updates}
+
+\item{default.assay}{Name of default assay for this object}
+
+\item{embeddings, loadings, projected, jackstraw}{Pull cell embeddings, feature
+loadings, projected loadings, and/or JackStraw information}
+}
+\value{
+\code{AssembleAssay}: An \code{Assay} object
+
+\code{AssembleDimReduc}: A \code{DimReduc} object
+
+\code{AssembleSeuratCommand}: A \code{SeuratCommand} object
+}
+\description{
+Assemble an object from an h5Seurat file
+}
+\keyword{internal}
diff --git a/man/AvailableData.Rd b/man/AvailableData.Rd
index 2526a4d..917f775 100644
--- a/man/AvailableData.Rd
+++ b/man/AvailableData.Rd
@@ -14,12 +14,15 @@ A dataframe with available Seurat datasets. Rownames of the dataframe are the ac
   \item{Installed}{Is the dataset installed?}
   \item{InstalledVersion}{Version of dataset installed, \code{NA} if not installed}
 }
-
-Other columns are extra metadata, and may change as new datasets are made available
 }
 \description{
 Get list of available datasets
+}
+\examples{
+AvailableData()
+
 }
 \seealso{
-\code{\link{InstallData}} \code{\link{InstalledData}} \code{\link{RemoveData}} \code{\link{UpdateData}}
+\code{\link{InstallData}} \code{\link{InstalledData}}
+\code{\link{RemoveData}} \code{\link{UpdateData}}
 }
diff --git a/man/Enumerate.Rd b/man/Enumerate.Rd
new file mode 100644
index 0000000..2e95974
--- /dev/null
+++ b/man/Enumerate.Rd
@@ -0,0 +1,28 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/zzz.R
+\name{Enumerate}
+\alias{Enumerate}
+\title{Enumerate a list or vector}
+\usage{
+Enumerate(x)
+}
+\arguments{
+\item{x}{A list or a vector}
+}
+\value{
+A list of length \code{x} with the following named values:
+\describe{
+  \item{\code{name}}{The name of \code{x} at a given index}
+  \item{\code{value}}{The value of \code{x} at the corresponding index}
+}
+}
+\description{
+Enumerate a list or vector
+}
+\note{
+For any given index \code{i} in \code{x}, all attempts to use the name
+of the value of \code{x} at \code{i} will be made. If there is no name
+(eg. \code{nchar(x = names(x = x)[i]) == 0}), the index \code{i} will be used
+in its stead
+}
+\keyword{internal}
diff --git a/man/GetObject.Rd b/man/GetObject.Rd
new file mode 100644
index 0000000..a077426
--- /dev/null
+++ b/man/GetObject.Rd
@@ -0,0 +1,49 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{GetObject}
+\alias{GetObject}
+\alias{GetAssays}
+\alias{GetDimReducs}
+\alias{GetGraphs}
+\title{Figure out which objects to load from an h5Seurat file}
+\usage{
+GetAssays(assays, index)
+
+GetDimReducs(reductions, index, assays = NULL)
+
+GetGraphs(index, assays = NULL)
+}
+\arguments{
+\item{assays}{One of:
+\itemize{
+  \item A character vector with names of assays
+  \item A character vector with one or more of \code{counts}, \code{data},
+  \code{scale.data} describing which slots of \strong{all assays} to load
+  \item A named list where each entry is either the name of an assay or a vector
+  describing which slots (described above) to take from which assay
+  \item \code{NULL} for all assays
+}}
+
+\item{index}{An h5Seurat index (\code{h5SI}) from \code{\link{IndexH5Seurat}}}
+
+\item{reductions}{A vector of reduction names to load}
+
+\item{graphs}{A vector of graph names to load}
+}
+\value{
+\code{GetAssays}: A named list where each entry is a vector describing
+the slots of an assay to load and the names are the assays to load
+
+\code{GetDimReducs}: A vector of reduction names that are derived from
+an assay in \code{assays}
+
+\code{GetGraphs}: A vector of graph names that are derived from an
+assay in \code{assays}
+}
+\description{
+Figure out which objects to load from an h5Seurat file
+}
+\seealso{
+\code{\link{LoadH5Seurat}} \code{\link{IndexH5Seurat}}
+}
+\keyword{internal}
diff --git a/man/IndexH5Seurat.Rd b/man/IndexH5Seurat.Rd
new file mode 100644
index 0000000..2283146
--- /dev/null
+++ b/man/IndexH5Seurat.Rd
@@ -0,0 +1,29 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{IndexH5Seurat}
+\alias{IndexH5Seurat}
+\alias{IndexH5Seurat.character}
+\alias{IndexH5Seurat.H5File}
+\title{Index an H5Seurat object}
+\usage{
+IndexH5Seurat(file, ...)
+
+\method{IndexH5Seurat}{character}(file, ...)
+
+\method{IndexH5Seurat}{H5File}(file, ...)
+}
+\arguments{
+\item{file}{Name of h5Seurat or connected h5Seurat file to load}
+
+\item{...}{Arguments passed to other methods}
+}
+\value{
+An object of class \code{h5SI}
+}
+\description{
+Index an H5Seurat object
+}
+\seealso{
+\code{\link{LoadH5Seurat}} \code{\link{SaveH5Seurat}}
+}
+\keyword{internal}
diff --git a/man/InstallData.Rd b/man/InstallData.Rd
index 5018550..cf79746 100644
--- a/man/InstallData.Rd
+++ b/man/InstallData.Rd
@@ -18,7 +18,14 @@ Invisible \code{NULL}
 }
 \description{
 Install a dataset
+}
+\examples{
+\dontrun{
+InstallData('pbmc3k')
+}
+
 }
 \seealso{
-\code{\link{AvailableData}} \code{\link{InstalledData}} \code{\link{RemoveData}} \code{\link{UpdateData}}
+\code{\link{AvailableData}} \code{\link{InstalledData}}
+\code{\link{RemoveData}} \code{\link{UpdateData}}
 }
diff --git a/man/InstalledData.Rd b/man/InstalledData.Rd
index 74852ce..24d6f89 100644
--- a/man/InstalledData.Rd
+++ b/man/InstalledData.Rd
@@ -19,7 +19,12 @@ Other columns are extra metadata, and may change as new datasets are made availa
 }
 \description{
 Get a list of installed datasets
+}
+\examples{
+InstalledData()
+
 }
 \seealso{
-\code{\link{AvailableData}} \code{\link{InstallData}} \code{\link{RemoveData}} \code{\link{UpdateData}}
+\code{\link{AvailableData}} \code{\link{InstallData}}
+\code{\link{RemoveData}} \code{\link{UpdateData}}
 }
diff --git a/man/IsMatrixEmpty.Rd b/man/IsMatrixEmpty.Rd
new file mode 100644
index 0000000..acd15c3
--- /dev/null
+++ b/man/IsMatrixEmpty.Rd
@@ -0,0 +1,37 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/zzz.R
+\name{IsMatrixEmpty}
+\alias{IsMatrixEmpty}
+\title{Check to see if a matrix is empty}
+\usage{
+IsMatrixEmpty(x)
+}
+\arguments{
+\item{x}{A matrix}
+}
+\value{
+\code{TRUE} if the matrix is empty otherwise \code{FALSE}
+}
+\description{
+Deterime if a matrix is empty or not
+}
+\details{
+A matrix is considered empty if it satisfies one of the following
+conditions:
+\itemize{
+  \item The dimensions of the matrix are 0-by-0 (\code{all(dim(x) == 0)})
+  \item The dimensions of the matrix are 1-by-1 (\code{all(dim(x) == 1)}) and
+  the sole value is a \code{NA}
+}
+These two situations correspond to matrices generated with either
+\code{new('matrix')} or \code{matrix()}
+}
+\examples{
+\dontrun{
+IsMatrixEmpty(new('matrix'))
+IsMatrixEmpty(matrix())
+IsMatrixEmpty(matrix(1:9, nrow = 3))
+}
+
+}
+\keyword{internal}
diff --git a/man/LoadData.Rd b/man/LoadData.Rd
new file mode 100644
index 0000000..35ccfe8
--- /dev/null
+++ b/man/LoadData.Rd
@@ -0,0 +1,73 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/seurat_data.R
+\name{LoadData}
+\alias{LoadData}
+\title{Modularly load a dataset}
+\usage{
+LoadData(ds, type = "info", assays = NULL, reductions = NULL,
+  graphs = NULL, object = NULL, overwrite = FALSE, verbose = TRUE)
+}
+\arguments{
+\item{ds}{Name of dataset to load}
+
+\item{type}{How to load the \code{Seurat} object; choose from
+\describe{
+  \item{info}{
+  Information about the object and what's stored in it; to specify a dataset
+  to get information for, pass 'info_\code{name of dataset}' based on the name
+  of the dataset in the \code{other.datasets} column of the data manifest
+  }
+  \item{raw}{The raw form of the dataset, no other options are evaluated}
+  \item{...}{
+  Name of dataset in the \code{other.datasets} column of the data manifest
+  corresponding to the dataset desired (eg. 'processed' for ifnb)
+  }
+}}
+
+\item{assays}{One of:
+\itemize{
+  \item A character vector with names of assays
+  \item A character vector with one or more of \code{counts}, \code{data},
+  \code{scale.data} describing which slots of \strong{all assays} to load
+  \item A named list where each entry is either the name of an assay or a vector
+  describing which slots (described above) to take from which assay
+  \item \code{NULL} for all assays
+}}
+
+\item{reductions}{One of:
+\itemize{
+  \item A character vector with names of reductions
+  \item \code{NULL} for all reductions
+  \item \code{NA} or \code{FALSE} for no reductions
+}
+\strong{Note}: Only reductions associated with an assay loaded in \code{assays}}
+
+\item{graphs}{One of:
+\itemize{
+  \item A character vector with names of graphs
+  \item \code{NULL} for all graphs
+  \item \code{NA} or \code{FALSE} for no reductions
+}}
+
+\item{object}{A \code{Seurat} object to append data to}
+
+\item{verbose}{Show progress updates}
+}
+\value{
+If \code{type} is info, information about the data contained within the
+h5Seurat file. Otherwise, a \code{Seurat} object with the data requested
+}
+\description{
+Modularly load a dataset
+}
+\examples{
+\dontrun{
+LoadData('ifnb')
+ifnb.raw <- LoadData('ifnb', type = 'raw')
+ifnb.proccessed <- LoadData('ifnb', type = 'processed')
+}
+
+}
+\seealso{
+\code{\link[utils]{data}}
+}
diff --git a/man/LoadH5Seurat.Rd b/man/LoadH5Seurat.Rd
new file mode 100644
index 0000000..f2531cc
--- /dev/null
+++ b/man/LoadH5Seurat.Rd
@@ -0,0 +1,69 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{LoadH5Seurat}
+\alias{LoadH5Seurat}
+\alias{LoadH5Seurat.character}
+\alias{LoadH5Seurat.H5File}
+\title{Load a saved \code{Seurat} object from an h5Seurat file}
+\usage{
+LoadH5Seurat(file, ...)
+
+\method{LoadH5Seurat}{character}(file, type = c("info", "object"),
+  assays = NULL, reductions = NULL, graphs = NULL, verbose = TRUE,
+  ...)
+
+\method{LoadH5Seurat}{H5File}(file, type = c("info", "object"),
+  assays = NULL, reductions = NULL, graphs = NULL, verbose = TRUE,
+  ...)
+}
+\arguments{
+\item{file}{Name of h5Seurat or connected h5Seurat file to load}
+
+\item{...}{Arguments passed to other methods}
+
+\item{assays}{One of:
+\itemize{
+  \item A character vector with names of assays
+  \item A character vector with one or more of \code{counts}, \code{data},
+  \code{scale.data} describing which slots of \strong{all assays} to load
+  \item A named list where each entry is either the name of an assay or a vector
+  describing which slots (described above) to take from which assay
+  \item \code{NULL} for all assays
+}}
+
+\item{reductions}{One of:
+\itemize{
+  \item A character vector with names of reductions
+  \item \code{NULL} for all reductions
+  \item \code{NA} or \code{FALSE} for no reductions
+}
+\strong{Note}: Only reductions associated with an assay loaded in \code{assays}}
+
+\item{graphs}{One of:
+\itemize{
+  \item A character vector with names of graphs
+  \item \code{NULL} for all graphs
+  \item \code{NA} or \code{FALSE} for no reductions
+}}
+
+\item{verbose}{Show progress updates}
+
+\item{meta.data}{Load object metadata?}
+
+\item{commands}{Load command information}
+
+\item{misc}{Load miscellaneous data?}
+
+\item{tools}{Load tool-specific information?}
+}
+\value{
+If \code{type} is info, information about the data contained within the
+h5Seurat file. Otherwise, a \code{Seurat} object with the data requested
+}
+\description{
+Load a saved \code{Seurat} object from an h5Seurat file
+}
+\seealso{
+\code{\link{SaveH5Seurat}} \code{\link{IndexH5Seurat}}
+}
+\keyword{internal}
diff --git a/man/MakeSpace.Rd b/man/MakeSpace.Rd
new file mode 100644
index 0000000..0f3a6eb
--- /dev/null
+++ b/man/MakeSpace.Rd
@@ -0,0 +1,24 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/zzz.R
+\name{MakeSpace}
+\alias{MakeSpace}
+\title{Make a space}
+\usage{
+MakeSpace(n)
+}
+\arguments{
+\item{n}{Length space should be}
+}
+\value{
+A space (' ') of length \code{n}
+}
+\description{
+Make a space
+}
+\examples{
+\dontrun{
+MakeSpace(10)
+}
+
+}
+\keyword{internal}
diff --git a/man/NameToPackage.Rd b/man/NameToPackage.Rd
index 92a3d14..af9d743 100644
--- a/man/NameToPackage.Rd
+++ b/man/NameToPackage.Rd
@@ -14,5 +14,13 @@ A vector of package names based on dataset names
 }
 \description{
 Find dataset packages based on name
+}
+\examples{
+\dontrun{
+NameToPackage('cbmc')
+NameToPackage('pbmc3k.SeuratData')
+NameToPackage('notadataset')
+}
+
 }
 \keyword{internal}
diff --git a/man/RemoveData.Rd b/man/RemoveData.Rd
index 2f5e21e..57433e5 100644
--- a/man/RemoveData.Rd
+++ b/man/RemoveData.Rd
@@ -15,7 +15,14 @@ RemoveData(ds, lib)
 }
 \description{
 Remove a dataset
+}
+\examples{
+\dontrun{
+RemoveData('pbmc3k')
+}
+
 }
 \seealso{
-\code{\link{AvailableData}} \code{\link{InstallData}} \code{\link{InstalledData}} \code{\link{UpdateData}}
+\code{\link{AvailableData}} \code{\link{InstallData}}
+\code{\link{InstalledData}} \code{\link{UpdateData}}
 }
diff --git a/man/SaveH5Seurat.Rd b/man/SaveH5Seurat.Rd
new file mode 100644
index 0000000..eccbdaf
--- /dev/null
+++ b/man/SaveH5Seurat.Rd
@@ -0,0 +1,37 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{SaveH5Seurat}
+\alias{SaveH5Seurat}
+\alias{SaveH5Seurat.default}
+\alias{SaveH5Seurat.Seurat}
+\title{Save a \code{Seurat} object to a h5Seurat file}
+\usage{
+SaveH5Seurat(object, filename, overwrite = FALSE, verbose = TRUE, ...)
+
+\method{SaveH5Seurat}{default}(object, filename, overwrite = FALSE,
+  verbose = TRUE, ...)
+
+\method{SaveH5Seurat}{Seurat}(object, filename = paste0(Project(object =
+  object), ".h5Seurat"), overwrite = FALSE, verbose = TRUE, ...)
+}
+\arguments{
+\item{object}{A \code{Seurat} object}
+
+\item{filename}{Name for h5Seurat file}
+
+\item{overwrite}{If an h5Seurat file of name \code{filename} exists, overwrite it?}
+
+\item{verbose}{Show progress updates}
+
+\item{...}{Arguments passed to other methods}
+}
+\value{
+Invisibly returns the name of the h5Seurat file
+}
+\description{
+Save a \code{Seurat} object to a h5Seurat file
+}
+\seealso{
+\code{\link{LoadH5Seurat}} \code{\link{IndexH5Seurat}}
+}
+\keyword{internal}
diff --git a/man/SeuratData-package.Rd b/man/SeuratData-package.Rd
index 9036e29..18498c2 100644
--- a/man/SeuratData-package.Rd
+++ b/man/SeuratData-package.Rd
@@ -14,11 +14,19 @@ Single cell RNA sequencing datasets can be large, consisting of matrices
 \section{Package options}{
 
 
-SeuratData uses the following [options()] to configure behaviour:
+SeuratData uses the following options to control behaviour, users can configure
+these with \code{\link[base]{options}}:
 
 \itemize{
   \item `SeuratData.repo.use`: Set the location where the SeuratData datasets
   are stored. Users generally should not modify.
+  \item `SeuratData.manifest.cache`: Cache the data manifest whenever we talk
+  to the data repository; note, setting to \code{FALSE} will simply prevent
+  SeuratData from caching the manifest, not from reading a previously cached
+  manifest
+  \item `SeuratData.roaming`: For Windows users, use a roaming profile directory
+  for domain users. See \url{https://en.wikipedia.org/wiki/Roaming_user_profile}
+  for a brief overview and Microsoft's documentation for greater detail
 }
 }
 
diff --git a/man/UpdateData.Rd b/man/UpdateData.Rd
index ccdf2d2..f389c41 100644
--- a/man/UpdateData.Rd
+++ b/man/UpdateData.Rd
@@ -25,7 +25,14 @@ Invisible \code{NULL}
 }
 \description{
 Update datasets
+}
+\examples{
+\dontrun{
+UpdateData(ask = FALSE)
+}
+
 }
 \seealso{
-\code{\link{AvailableData}} \code{\link{InstallData}} \code{\link{InstalledData}} \code{\link{RemoveData}}
+\code{\link{AvailableData}} \code{\link{InstallData}}
+\code{\link{InstalledData}} \code{\link{RemoveData}}
 }
diff --git a/man/ValidateH5SI.Rd b/man/ValidateH5SI.Rd
new file mode 100644
index 0000000..f0c375a
--- /dev/null
+++ b/man/ValidateH5SI.Rd
@@ -0,0 +1,18 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{ValidateH5SI}
+\alias{ValidateH5SI}
+\title{Validate an h5Seurat Index}
+\usage{
+ValidateH5SI(x)
+}
+\arguments{
+\item{x}{An h5Seurat Index (h5SI)}
+}
+\value{
+\code{x} if valid, otherwise a modified \code{x} to be valid
+}
+\description{
+Validate an h5Seurat Index
+}
+\keyword{internal}
diff --git a/man/WriteH5List.Rd b/man/WriteH5List.Rd
new file mode 100644
index 0000000..37fa7f3
--- /dev/null
+++ b/man/WriteH5List.Rd
@@ -0,0 +1,23 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{WriteH5List}
+\alias{WriteH5List}
+\title{Write a list as a series of HDF5 groups and datasets}
+\usage{
+WriteH5List(x, name, hgroup)
+}
+\arguments{
+\item{x}{A list}
+
+\item{name}{Name to save dataset as}
+
+\item{hgroup}{An HDF5 file or group (\code{H5File} or
+\code{H5Group} objects from hdf5r)}
+}
+\value{
+Invisibly returns \code{NULL}
+}
+\description{
+Write a list as a series of HDF5 groups and datasets
+}
+\keyword{internal}
diff --git a/man/as.list.H5Group.Rd b/man/as.list.H5Group.Rd
new file mode 100644
index 0000000..66a2c7c
--- /dev/null
+++ b/man/as.list.H5Group.Rd
@@ -0,0 +1,23 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{as.list.H5Group}
+\alias{as.list.H5Group}
+\alias{as.list}
+\title{Read an HDF5 group into a ist}
+\usage{
+\method{as.list}{H5Group}(x, ...)
+}
+\arguments{
+\item{x}{An \code{H5Group} object (from hdf5r)}
+
+\item{...}{Ignored}
+}
+\value{
+A list with the data contained within \code{x}; if the HDF5 attribute
+\code{s4class} is set and is a class, will return an object of class \code{s4class}
+instead
+}
+\description{
+Read an HDF5 group into a ist
+}
+\keyword{internal}
diff --git a/man/h5SI.Rd b/man/h5SI.Rd
new file mode 100644
index 0000000..d94b2a1
--- /dev/null
+++ b/man/h5SI.Rd
@@ -0,0 +1,18 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{h5SI}
+\alias{h5SI}
+\title{Construct an h5Seurat index}
+\usage{
+h5SI(...)
+}
+\arguments{
+\item{...}{...}
+}
+\value{
+...
+}
+\description{
+Construct an h5Seurat index
+}
+\keyword{internal}
diff --git a/man/print.h5SI.Rd b/man/print.h5SI.Rd
new file mode 100644
index 0000000..a072321
--- /dev/null
+++ b/man/print.h5SI.Rd
@@ -0,0 +1,21 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/h5seurat.R
+\name{print.h5SI}
+\alias{print.h5SI}
+\alias{print}
+\title{Print index information for an h5Seurat Index (h5SI)}
+\usage{
+\method{print}{h5SI}(x, ...)
+}
+\arguments{
+\item{x}{an object used to select a method.}
+
+\item{...}{further arguments passed to or from other methods.}
+}
+\value{
+Invisibly returns \code{x}
+}
+\description{
+Print index information for an h5Seurat Index (h5SI)
+}
+\keyword{internal}
diff --git a/man/set-if-null.Rd b/man/set-if-null.Rd
index 4db00db..0de9104 100644
--- a/man/set-if-null.Rd
+++ b/man/set-if-null.Rd
@@ -17,6 +17,13 @@ rhs if lhs is null, else lhs
 }
 \description{
 Set a default value if an object is null
+}
+\examples{
+\dontrun{
+4 \%||\% 5
+NULL \%|| 5
+}
+
 }
 \references{
 \url{https://adv-r.hadley.nz/functions.html#missing-arguments}