biopipen 0.21.0__py3-none-any.whl → 0.34.26__py3-none-any.whl

biopipen/scripts/scrna/SeuratMap2Ref.R

@@ -1,139 +1,110 @@
-source("{{biopipen_dir}}/utils/misc.R")
-
 library(Seurat)
-library(SeuratDisk)
+library(rlang)
+library(biopipen.utils)
+
+set.seed(8525)
 
 sobjfile = {{in.sobjfile | r}}
 outfile = {{out.outfile | r}}
+joboutdir = {{job.outdir | r}}
 use = {{envs.use | r}}
-alias = {{envs.alias | r}}
+ident = {{envs.ident | r}}
 ref = {{envs.ref | r}}
+refnorm = {{envs.refnorm | r}}
+ncores = {{envs.ncores | r}}
+split_by = {{envs.split_by | r}}
+mutaters = {{envs.mutaters | r}}
+skip_if_normalized = {{envs.skip_if_normalized | r}}
 sctransform_args = {{envs.SCTransform | r: todot="-"}}
+normalizedata_args = {{envs.NormalizeData | r: todot="-"}}
 findtransferanchors_args = {{envs.FindTransferAnchors | r: todot="-"}}
 mapquery_args = {{envs.MapQuery | r: todot="-"}}
-mappingscore_args = {{envs.MappingScore | r: todot="-"}}
+cache = {{envs.cache | r}}
+plots = {{envs.plots | r}}
 
-outdir = dirname(outfile)
+log <- get_logger()
+reporter <- get_reporter()
+
+options(future.globals.maxSize = Inf)
+options(future.rng.onMisuse="ignore")
+options(Seurat.object.assay.version = "v5")
 
 # See if we have a reference
 if (is.null(ref)) {
-  stop("No reference provided")
+    stop("No reference provided (envs.ref)")
 }
 
-.expand_dims = function(args, name = "dims") {
-    # Expand dims from 30 to 1:30
-    if (is.numeric(args[[name]]) && length(args[[name]] == 1)) {
-        args[[name]] = 1:args[[name]]
-    }
-    args
-}
-findtransferanchors_args = .expand_dims(findtransferanchors_args)
-
-# Load reference
-print("Loading reference")
-if (endsWith(ref, ".rds") || endsWith(ref, ".RDS")) {
-    reference = readRDS(ref)
-} else if (endsWith(ref, ".h5ad") || endsWith(ref, ".H5AD")) {
-    reference = ReadH5AD(ref)
-} else {
-    reference = LoadH5Seurat(ref)
+if (is.null(use)) {
+    stop("No use provided (envs.use), don't know which column to transfer as cluster")
 }
 
-# Load Seurat object
-print("Loading Seurat object")
-sobj = readRDS(sobjfile)
-
-# Normalize data
-print("Normalizing data")
-sctransform_args$object = sobj
-sctransform_args$residual.features = rownames(x = reference)
-query = do_call(SCTransform, sctransform_args)
-
-# Find anchors between query and reference
-print("Finding anchors")
-findtransferanchors_args$reference = reference
-findtransferanchors_args$query = query
-anchors = do_call(FindTransferAnchors, findtransferanchors_args)
-
-# Map query to reference
-print("Mapping query to reference")
-mapquery_args$reference = reference
-mapquery_args$query = query
-mapquery_args$anchorset = anchors
-query = do_call(MapQuery, mapquery_args)
-
-# Calculating mapping score
-print("Calculating mapping score")
-mappingscore_args$anchors = anchors
-mappingscore = tryCatch({
-    do_call(MappingScore, mappingscore_args)
-}, error = function(e) {
-    if (e$message == "subscript out of bounds") {
-        stop(paste0(
-            "While calculating mapping score, the following error was encountered: \n",
-            "subscript out of bounds.  \n\n",
-            "You may want to try a smaller `ndim` (default: 50) in `envs.MappingScore`."
-        ))
-    }
-    stop(e)
-})
-
-# Calculate mapping score and add to metadata
-print("Calculating mapping score")
-query = AddMetaData(
-  object = query,
-  metadata = mappingscore,
-  col.name = "mapping.score"
-)
+outdir = dirname(outfile)
+if (isTRUE(cache)) {
+    cache = joboutdir
+}
+if (is.null(split_by)) {
+    future::plan(strategy = "multicore", workers = ncores)
+}
 
-# Add the alias to the metadata for the clusters
-print("Adding alias to metadata")
-query = AddMetaData(
-  object = query,
-  metadata = query[[use]],
-  col.name = alias
+log$info("Loading reference ...")
+if (endsWith(ref, ".rds") || endsWith(ref, ".RDS") || endsWith(ref, ".qs") || endsWith(ref, ".qs2")) {
+    reference <- read_obj(ref)
+} else if (endsWith(ref, ".h5seurat") || endsWith(ref, ".H5Seurat")) {
+    reference <- SeuratDisk::LoadH5Seurat(ref)
+} else {
+    stop("Reference file must be .qs, .qs2, .rds, .RDS, .h5seurat or .H5Seurat")
+}
+reference <- tryCatch(JoinLayers(reference), error = function(e) {reference})
+Idents(reference) <- reference@meta.data[[use]]
+
+log$info("Loading query data ...")
+sobj <- read_obj(sobjfile)
+
+sobj <- RunSeuratMap2Ref(
+    object = sobj, ref = reference, use = use,
+    ident = ident, refnorm = refnorm, skip_if_normalized = skip_if_normalized,
+    split_by = split_by, ncores = ncores,
+    SCTransformArgs = sctransform_args,
+    NormalizeDataArgs = normalizedata_args,
+    FindTransferAnchorsArgs = findtransferanchors_args,
+    MapQueryArgs = mapquery_args,
+    log = log, cache = cache
 )
 
 # Save
-print("Saving")
-saveRDS(query, file = outfile)
+gc()
+log$info("Saving result ...")
+save_obj(sobj, file = outfile)
 
 
-# ############################
-# Some plots
-# ############################
-
-# # Plot the UMAP
-print("Plotting")
-for (refname in names(mapquery_args$refdata)) {
-    if (refname == "predicted_ADT") {
+### Plotting
+log$info("Plotting features ...")
+for (name in names(plots)) {
+    if (is.null(plots[[name]])) {
         next
     }
-    reduction = if (is.null(mapquery_args$reduction.model)) "wnn.umap" else mapquery_args$reduction.model
-    print(paste0("Plotting UMAP for ", refname))
-    p = DimPlot(
-        object = reference,
-        reduction = reduction,
-        group.by = refname,
-        label = TRUE,
-        label.size = 3,
-        repel = TRUE,
-    ) + NoLegend()
-
-    png(file.path(outdir, paste0("Reference_UMAP_", refname, ".png")), width = 1000, height = 1000, res = 100)
-    print(p)
-    dev.off()
-
-    p = DimPlot(
-        object = query,
-        reduction = "ref.umap",
-        group.by = paste0("predicted.", refname),
-        label = TRUE,
-        label.size = 3,
-        repel = TRUE,
-    ) + NoLegend()
-
-    png(file.path(outdir, paste0("Query_UMAP_", refname, ".png")), width = 1000, height = 1000, res = 100)
-    print(p)
-    dev.off()
+    log$info("- {name} ...")
+    plots[[name]]$features <- gsub("{use}", use, plots[[name]]$features, fixed = TRUE)
+    plots[[name]]$features <- gsub("{ident}", ident, plots[[name]]$features, fixed = TRUE)
+
+    plots[[name]]$devpars <- plots[[name]]$devpars %||% list()
+    plots[[name]]$devpars$res <- plots[[name]]$devpars$res %||% 100
+    plots[[name]]$devpars$width <- plots[[name]]$devpars$width %||% 1200
+    plots[[name]]$devpars$height <- plots[[name]]$devpars$height %||% 720
+    plots[[name]]$more_formats <- plots[[name]]$more_formats %||% character()
+    plots[[name]]$save_code <- FALSE
+    plots[[name]]$descr <- plots[[name]]$descr %||% name
+    extract_vars(plots[[name]], "devpars", "more_formats", "save_code", "descr")
+
+    plot_fn <- gglogger::register(VizSeuratMap2Ref)
+    p <- do_call(plot_fn, c(list(query = sobj, ref = reference), plots[[name]]))
+    prefix <- file.path(outdir, paste0(slugify(name), ".map2ref"))
+    save_plot(p, prefix, devpars, formats = c("png", more_formats))
+
+    reporter$add(
+        reporter$image(prefix, more_formats, save_code = FALSE, kind = "image"),
+        h1 = name
+    )
 }
+
+reporter$save(joboutdir)

biopipen/scripts/scrna/SeuratMetadataMutater.R

@@ -1,20 +1,30 @@
-source("{{biopipen_dir}}/utils/mutate_helpers.R")
 library(rlang)
 library(tibble)
 library(dplyr)
 library(Seurat)
+library(biopipen.utils)
 
-srtobj = {{in.srtobj | quote}}
+srtobj = {{in.srtobj | r}}
 metafile = {{in.metafile | r}}
 mutaters = {{envs.mutaters | r}}
-rdsfile = {{out.rdsfile | quote}}
+outfile = {{out.outfile | r}}
 
-srt = readRDS(srtobj)
+srt = read_obj(srtobj)
 metadata = srt@meta.data
 
 if (!is.null(metafile)) {
     mdata = read.table(metafile, header=TRUE, row.names=1, sep="\t", check.names=FALSE)
-    metadata = cbind(metadata, mdata[rownames(metadata),,drop=FALSE])
+    ov_cols = intersect(colnames(metadata), colnames(mdata))
+    if (length(ov_cols) > 0) {
+        log_warn(paste0(
+            "The following columns are already present in Seurat object and will be ignored: ",
+            paste(ov_cols, collapse=', ')
+        ))
+    }
+    metadata = cbind(
+        metadata,
+        mdata[rownames(metadata), setdiff(colnames(mdata), ov_cols), drop=FALSE]
+    )
 }
 
 expr = list()
@@ -28,4 +38,4 @@ if (!is.null(expr) && length(expr) > 0) {
     srt@meta.data = metadata
 }
 
-saveRDS(srt, rdsfile)
+save_obj(srt, outfile)