biopipen 0.33.1__py3-none-any.whl → 0.34.0__py3-none-any.whl

biopipen/scripts/scrna/SeuratClusterStats-features.R

@@ -60,6 +60,7 @@ do_one_features <- function(name) {
     log$info("- Case: {name}")
 
     case <- list_update(features_defaults, features[[name]])
+    case$descr <- case$descr %||% ""
     case <- extract_vars(
         case,
         "devpars", "more_formats", "save_code", "save_data", "order_by",
@@ -87,50 +88,58 @@ do_one_features <- function(name) {
     }
 
     info <- case_info(name, odir, is_dir = FALSE, create = TRUE)
-    case$features <- .get_features(features, case$object)
-    p <- do_call(gglogger::register(FeatureStatPlot), case)
-    save_plot(p, info$prefix, devpars, formats = c("png", more_formats))
-    if (save_code) {
-        save_plotcode(p, info$prefix,
-            setup = c("library(scplotter)", "load('data.RData')", "invisible(list2env('case'))"),
-            "case",
-            auto_data_setup = FALSE)
+
+    caching <- Cache$new(
+        c(case, list(devpars, more_formats, save_code, save_data, order_by, subset, features, descr)),
+        prefix = "biopipen.scrna.SeuratClusterStats.features",
+        cache_dir = cache,
+        kind = "prefix",
+        path = info$prefix
+    )
+
+    if (caching$is_cached()) {
+        log$info("  plots are cached, restoring ...")
+        caching$restore()
+    } else {
+        case$features <- .get_features(features, case$object)
+        p <- do_call(gglogger::register(FeatureStatPlot), case)
+        save_plot(p, info$prefix, devpars, formats = c("png", more_formats))
+        if (save_code) {
+            save_plotcode(p, info$prefix,
+                setup = c("library(scplotter)", "load('data.RData')", "invisible(list2env(case, envir = .GlobalEnv))"),
+                "case",
+                auto_data_setup = FALSE)
+        }
+
+        if (save_data) {
+            pdata <- attr(p, "data") %||% p$data
+            if (!inherits(pdata, "data.frame") && !inherits(pdata, "matrix")) {
+                stop("'save_data = TRUE' is not supported for plot_type: ", case$plot_type)
+            }
+            write.table(pdata, paste0(info$prefix, ".data.txt"), sep="\t", quote=FALSE, row.names=FALSE)
+        }
+
+        caching$save(info$prefix)
     }
-    if (exists("descr") && !is.null(descr)) {
+    # add reports
+    if (!is.null(descr) && nchar(descr) > 0) {
         reporter$add2(
-            list(
-                kind = "descr",
-                content = descr
-            ),
+            list(kind = "descr", content = descr),
             hs = c(info$section, info$name)
         )
     }
 
     if (save_data) {
-        if (!inherits(p$data, "data.frame") && !inherits(p$data, "matrix")) {
-            stop("'save_data = TRUE' is not supported for plot_type: ", case$plot_type)
-        }
-        write.table(p$data, paste0(info$prefix, ".data.txt"), sep="\t", quote=FALSE, row.names=FALSE)
         reporter$add2(
             list(
                 name = "Plot",
-                contents = list(
-                    reporter$image(
-                        info$prefix, more_formats, save_code, kind = "image")
-                )
+                contents = list(reporter$image(info$prefix, more_formats, save_code, kind = "image"))
             ),
             list(
                 name = "Data",
                 contents = list(
-                    list(
-                        kind = "descr",
-                        content = "Data used directly for the plot"
-                    ),
-                    list(
-                        kind = "table",
-                        src = paste0(info$prefix, ".data.txt"),
-                        data = list(nrows = 100)
-                    )
+                    list(kind = "descr", content = "Data used directly for the plot"),
+                    list(kind = "table", src = paste0(info$prefix, ".data.txt"), data = list(nrows = 100))
                 )
             ),
             hs = c(info$section, info$name),

biopipen/scripts/scrna/SeuratClusterStats-ngenes.R

@@ -12,79 +12,70 @@ do_one_ngenes <- function(name) {
 
     case <- list_update(ngenes_defaults, ngenes[[name]])
     case$devpars <- list_update(ngenes_defaults$devpars, case$devpars)
+    case$more_formats <- case$more_formats %||% character(0)
+    case$save_code <- case$save_code %||% FALSE
+    case$descr <- case$descr %||% name
+    case$save_data <- case$save_data %||% FALSE
+    case$ylab <- case$ylab %||% "Number of expressed genes"
+    case$features <- "Number of expressed genes"
+    extract_vars(case, "devpars", "more_formats", "descr", "save_code", "save_data", subset_ = "subset")
 
     if (!is.null(case$subset)) {
-        sobj <- srtobj %>% filter(!!rlang::parse_expr(case$subset))
+        case$object <- srtobj %>% filter(!!rlang::parse_expr(subset_))
     } else {
-        sobj <- srtobj
+        case$object <- srtobj
     }
-    df_cells <- sobj@meta.data %>% mutate(.nexpr = colSums(GetAssayData(sobj) > 0))
+    case$object <- AddMetaData(case$object, Matrix::colSums(GetAssayData(case$object) > 0), col.name = "Number of expressed genes")
 
-    select_cols = c(case$ident, case$group.by, case$split.by, ".nexpr")
-    df_cells = df_cells %>% select(all_of(select_cols))
-
-    p = df_cells |>
-        ggplot(aes(
-            x=!!sym(case$ident),
-            y=.nexpr,
-            fill=!!sym(ifelse(is.null(case$group.by), case$ident, case$group.by))
-        )) +
-        geom_violin(alpha = 0.6, position = ifelse(is.null(case$group.by), "identity", "dodge")) +
-        geom_boxplot(
-            position = ifelse(is.null(case$group.by), "identity", "dodge"),
-            width = .1,
-            fill = "white"
-        ) +
-        plotthis::theme_this() +
-        # scale_fill_biopipen() +
-        ylab("Number of genes expressed")
-
-    if (!is.null(case$split.by)) {
-        p = p + facet_wrap(case$split.by)
+    info <- case_info(name, odir, is_dir = FALSE, create = TRUE)
+    p <- do_call(gglogger::register(FeatureStatPlot), case)
+    save_plot(p, info$prefix, case$devpars, formats = c("png", more_formats))
+    if (save_code) {
+        save_plotcode(p, info$prefix,
+            setup = c("library(scplotter)", "load('data.RData')", "invisible(list2env(case, envir = .GlobalEnv))"),
+            "case",
+            auto_data_setup = FALSE
+        )
     }
-
-    figprefix = file.path(odir, paste0(slugify(name), ".boxplot"))
-
-    save_plot(p, figprefix, case$devpars)
-    save_plotcode(
-        p,
-        figprefix,
-        c(
-            'library(rlang)',
-            'library(ggplot2)',
-            'library(ggprism)',
-            '',
-            'load("data.RData")'
-        ),
-        "df_cells", "case"
-    )
-
-    reporter$add(
-        list(
-            kind = "descr",
-            content = paste0(
-                "Plots showing the number of genes expressed in each ",
-                case$ident,
-                ifelse(
-                    is.null(case$group.by),
-                    "",
-                    paste0(", by ", paste0(case$group.by, collapse = ", "))
+    if (save_data) {
+        pdata <- attr(p, "data") %||% p$data
+        if (!inherits(pdata, "data.frame") && !inherits(pdata, "matrix")) {
+            stop("'save_data = TRUE' is not supported for plot_type: ", case$plot_type)
+        }
+        write.table(pdata, paste0(info$prefix, ".data.txt"), sep = "\t", quote = FALSE, row.names = FALSE)
+        reporter$add2(
+            list(
+                name = "Plot",
+                contents = list(
+                    list(kind = "descr", content = case$descr),
+                    reporter$image(info$prefix, more_formats, save_code, kind = "image")
                 )
-            )
-        ),
-        list(
-            kind = "image",
-            src = paste0(figprefix, ".png"),
-            download = list(
-                paste0(figprefix, ".pdf"),
-                list(
-                    src = paste0(figprefix, ".code.zip"),
-                    tip = "Download the code to reproduce the plot",
-                    icon = "Code"
+            ),
+            list(
+                name = "Data",
+                contents = list(
+                    list(
+                        kind = "descr",
+                        content = "Data used directly for the plot"
+                    ),
+                    list(
+                        kind = "table",
+                        src = paste0(info$prefix, ".data.txt"),
+                        data = list(nrows = 100)
+                    )
                 )
-            )),
-        h1 = name
-    )
+            ),
+            hs = c(info$section, info$name),
+            ui = "tabs"
+        )
+    }
+    else {
+        reporter$add2(
+            list(kind = "descr", content = case$descr),
+            reporter$image(info$prefix, more_formats, save_code, kind = "image"),
+            hs = c(info$section, info$name)
+        )
+    }
 }
 
 sapply(names(ngenes), do_one_ngenes)

biopipen/scripts/scrna/SeuratClusterStats-stats.R

@@ -1,5 +1,4 @@
 # Loaded variables: srtfile, outdir, srtobj
-library(circlize)
 
 log$info("stats:")
 
@@ -23,16 +22,17 @@ do_one_stats <- function(name) {
     save_plot(p, info$prefix, devpars, formats = c("png", more_formats))
     if (save_code) {
         save_plotcode(p, info$prefix,
-            setup = c("library(scplotter)", "load('data.RData')", "invisible(list2env('case'))"),
+            setup = c("library(scplotter)", "load('data.RData')", "invisible(list2env(case, envir = .GlobalEnv))"),
             "case",
             auto_data_setup = FALSE)
     }
 
     if (save_data) {
-        if (!inherits(p$data, "data.frame") && !inherits(p$data, "matrix")) {
+        pdata <- attr(p, "data") %||% p$data
+        if (!inherits(pdata, "data.frame") && !inherits(pdata, "matrix")) {
             stop("'save_data = TRUE' is not supported for plot_type: ", case$plot_type)
         }
-        write.table(p$data, paste0(info$prefix, ".data.txt"), sep="\t", quote=FALSE, row.names=FALSE)
+        write.table(pdata, paste0(info$prefix, ".data.txt"), sep="\t", quote=FALSE, row.names=FALSE)
         reporter$add2(
             list(
                 name = "Plot",

biopipen/scripts/scrna/SeuratClusterStats.R

@@ -16,9 +16,12 @@ srtfile <- {{in.srtobj | r}}
 outdir <- {{out.outdir | r}}
 joboutdir <- {{job.outdir | r}}
 mutaters <- {{envs.mutaters | r}}
+cache <- {{envs.cache | r}}
+
+if (isTRUE(cache)) { cache = joboutdir }
 
 log$info("Loading Seurat object ...")
-srtobj = readRDS(srtfile)
+srtobj = read_obj(srtfile)
 
 log$info("Applying mutaters if any ...")
 if (!is.null(mutaters) && length(mutaters) > 0) {
@@ -29,26 +32,26 @@ if (!is.null(mutaters) && length(mutaters) > 0) {
 ############## clustree ##############
 clustrees_defaults <- {{envs.clustrees_defaults | r: todot="-"}}
 clustrees <- {{envs.clustrees | r: todot="-", skip=1}}
-{{ biopipen_dir | joinpaths: "scripts", "scrna", "SeuratClusterStats-clustree.R" | source_r }}
+{% include biopipen_dir + "/scripts/scrna/SeuratClusterStats-clustree.R" %}
 
 ############## stats ##############
 stats_defaults = {{envs.stats_defaults | r: todot="-"}}
 stats = {{envs.stats | r: todot="-", skip=1}}
-{{ biopipen_dir | joinpaths: "scripts", "scrna", "SeuratClusterStats-stats.R" | source_r }}
+{% include biopipen_dir + "/scripts/scrna/SeuratClusterStats-stats.R" %}
 
 ############## ngenes ##############
 ngenes_defaults <- {{envs.ngenes_defaults | r: todot="-"}}
 ngenes <- {{envs.ngenes | r: todot="-", skip=1}}
-{{ biopipen_dir | joinpaths: "scripts", "scrna", "SeuratClusterStats-ngenes.R" | source_r }}
+{% include biopipen_dir + "/scripts/scrna/SeuratClusterStats-ngenes.R" %}
 
 ############## features ##############
 features_defaults = {{envs.features_defaults | r: todot="-"}}
 features = {{envs.features | r: todot="-", skip=1}}
-{{ biopipen_dir | joinpaths: "scripts", "scrna", "SeuratClusterStats-features.R" | source_r: chdir = False }}
+{% include biopipen_dir + "/scripts/scrna/SeuratClusterStats-features.R" %}
 
 ############## dimplots ##############
 dimplots_defaults = {{envs.dimplots_defaults | r: todot="-"}}
 dimplots = {{envs.dimplots | r: todot="-", skip=1}}
-{{ biopipen_dir | joinpaths: "scripts", "scrna", "SeuratClusterStats-dimplots.R" | source_r }}
+{% include biopipen_dir + "/scripts/scrna/SeuratClusterStats-dimplots.R" %}
 
 reporter$save(joboutdir)

biopipen/scripts/scrna/SeuratClustering.R

@@ -1,56 +1,39 @@
-{{ biopipen_dir | joinpaths: "utils", "misc.R" | source_r }}
-{{ biopipen_dir | joinpaths: "utils", "caching.R" | source_r }}
 
 library(Seurat)
-library(future)
-library(rlang)
-library(tidyr)
-library(dplyr)
-library(digest)
+library(biopipen.utils)
 
 set.seed(8525)
 
-srtfile <- {{in.srtobj | quote}}
-rdsfile <- {{out.rdsfile | quote}}
-joboutdir <- {{job.outdir | quote}}
-envs <- {{envs | r: todot="-"}}
+srtfile <- {{in.srtobj | r}}
+outfile <- {{out.outfile | r}}
+joboutdir <- {{job.outdir | r}}
+RunPCAArgs <- {{envs.RunPCA | r: todot="-"}}
+FindNeighborsArgs <- {{envs.FindNeighbors | r: todot="-"}}
+FindClustersArgs <- {{envs.FindClusters | r: todot="-"}}
+RunUMAPArgs <- {{envs.RunUMAP | r: todot="-"}}
+cache <- {{envs.cache | r}}
+ncores <- {{envs.ncores | r}}
 
-if (length(envs$ScaleData) > 0 && length(envs$SCTransform) > 0) {
-    stop("Cannot specify both ScaleData and SCTransform")
-}
+log <- get_logger()
 
-options(str = strOptions(vec.len = 5, digits.d = 5))
+# options(str = strOptions(vec.len = 5, digits.d = 5))
 options(future.globals.maxSize = 80000 * 1024^2)
-plan(strategy = "multicore", workers = envs$ncores)
-
-{{ biopipen_dir | joinpaths: "scripts", "scrna", "SeuratClustering-common.R" | source_r }}
-
-envs$RunUMAP <- expand_dims(envs$RunUMAP)
-envs$FindNeighbors <- expand_dims(envs$FindNeighbors)
-
-log_info("Reading Seurat object ...")
-sobj <- readRDS(srtfile)
-
-if (isTRUE(envs$cache)) { envs$cache <- joboutdir }
-if (length(envs$cache) > 1) {
-    log_warn("Multiple cache directories (envs.cache) detected, using the first one.")
-    envs$cache <- envs$cache[1]
-}
-sobj_sig <- capture.output(str(sobj))
-dig_sig <- digest::digest(sobj_sig, algo = "md5")
-dig_sig <- substr(dig_sig, 1, 8)
-cache_dir <- NULL
-if (is.character(envs$cache)) {
-    cache_dir <- file.path(envs$cache, paste0(dig_sig, ".seurat_cache"))
-    dir.create(cache_dir, recursive = TRUE, showWarnings = FALSE)
-    writeLines(sobj_sig, file.path(cache_dir, "signature.txt"))
-}
-
-sobj <- run_transformation(sobj)
-sobj <- run_umap(sobj)
-sobj <- run_findneighbors(sobj)
-sobj <- run_findclusters(sobj)
-sobj <- run_prepsctfindmarkers(sobj)
-
-log_info("Saving results ...")
-saveRDS(sobj, file = rdsfile)
+plan(strategy = "multicore", workers = ncores)
+
+log$info("Reading Seurat object ...")
+sobj <- read_obj(srtfile)
+
+if (isTRUE(cache)) { cache <- joboutdir }
+
+sobj <- RunSeuratClustering(
+    sobj,
+    RunPCAArgs = RunPCAArgs,
+    RunUMAPArgs = RunUMAPArgs,
+    FindNeighborsArgs = FindNeighborsArgs,
+    FindClustersArgs = FindClustersArgs,
+    log = log,
+    cache = cache
+)
+
+log$info("Saving results ...")
+save_obj(sobj, file = outfile)

biopipen/scripts/scrna/SeuratLoading.R

@@ -1,7 +1,7 @@
 library(Seurat)
 
-metafile = {{in.metafile | quote}}
-rdsfile = {{out.rdsfile | quote}}
+metafile = {{in.metafile | r}}
+rdsfile = {{out.rdsfile | r}}
 
 metadata = read.table(
     metafile,