biopipen 0.32.1__py3-none-any.whl → 0.32.3__py3-none-any.whl

biopipen/__init__.py

@@ -1 +1 @@
-__version__ = "0.32.1"
+__version__ = "0.32.3"

biopipen/core/filters.py

@@ -15,7 +15,7 @@ filtermanager = FilterManager()
 @filtermanager.register
 def dict_to_cli_args(
     dic: Mapping[str, Any],
-    exclude: List[str] = None,
+    exclude: List[str] | None = None,
     prefix: str | None = None,
     sep: str | None = " ",
     dup_key: bool = True,
@@ -118,7 +118,7 @@ def dict_to_cli_args(
 def r(
     obj: Any,
     ignoreintkey: bool = True,
-    todot: str = None,
+    todot: str | None = None,
     sortkeys: bool = False,
     skip: int = 0,
     _i: int = 0,
@@ -271,9 +271,22 @@ def _render_fgsea(
                     "title": "Plot",
                     "ui": "flat",
                     "contents": [
+                        {
+                            "kind": "descr",
+                            "content": (
+                                "This table presents a comprehensive summary of the "
+                                "top enriched pathways derived from the fgsea. "
+                                "Each row corresponds to a pathway, and the gene ranks "
+                                "are shown based on the ranking metric used in the "
+                                "analysis. The enrichment score, p-value, and adjusted "
+                                "p-value are also provided to assess the significance "
+                                "of the enrichment."
+                            )
+                        },
                         {
                             "kind": "image",
                             "src": str(Path(cont["dir"]).joinpath("gsea_table.png")),
+                            "download": str(Path(cont["dir"]).joinpath("gsea_table.pdf"))
                         }
                     ],
                 },
@@ -281,6 +294,19 @@ def _render_fgsea(
                     "title": "Table",
                     "ui": "flat",
                     "contents": [
+                        {
+                            "kind": "descr",
+                            "content": (
+                                "This plot represents the GSEA results for a specified "
+                                "gene set, illustrating the distribution and impact of "
+                                "the gene set along the ranked list of genes. "
+                                "The running enrichment score curve shows the "
+                                "cumulative enrichment score as genes from the input "
+                                "list are encountered. Positive peaks on the curve "
+                                "indicate regions where members of the gene set are "
+                                "predominantly found."
+                            )
+                        },
                         {
                             "kind": "table",
                             "src": str(Path(cont["dir"]).joinpath("fgsea.txt")),
@@ -297,6 +323,7 @@ def _render_fgsea(
             "contents": [
                 {
                     "src": str(Path(cont["dir"]) / f"fgsea_{slug}.png"),
+                    "download": str(Path(cont["dir"]) / f"fgsea_{slug}.pdf"),
                     "title": pw,
                 }
                 for pw, slug in pathways
@@ -304,7 +331,7 @@ def _render_fgsea(
         },
     ]
 
-    return render_ui(components, "accordion", job, level)
+    return render_ui(components, "accordion", job, level)  # type: ignore
 
 
 @register_component("pdf")
@@ -359,11 +386,23 @@ def _render_enrichr(
                             "title": "Plot",
                             "ui": "flat",
                             "contents": [
+                                {
+                                    "kind": "descr",
+                                    "content": (
+                                        "The enrichment analysis results of the top "
+                                        "biological pathways associated with the input "
+                                        "gene set. Each bar represents a pathway, "
+                                        "with the length of the bar indicating the "
+                                        "number of input genes overlapping with genes "
+                                        "in that pathway. The color intensity of the "
+                                        "bars reflects the statistical significance of "
+                                        "the enrichment (p-value). "
+                                    )
+                                },
                                 {
                                     "kind": "image",
-                                    "src": str(
-                                        Path(cont["dir"]).joinpath(f"Enrichr-{db}.png")
-                                    ),
+                                    "src": str(enrichr_plot),
+                                    "download": str(enrichr_plot.with_suffix(".pdf")),
                                 }
                             ],
                         },

biopipen/reports/utils/misc.liq

@@ -43,10 +43,12 @@ import { Image } from "$libs";
                 {{ caps[i] }}
             </div>
             {% endif %}
-            <Image
-                style="max-width: 90%"
-                src={{src | quote}}
-            />
+            {% if src | replace: ".png", ".pdf" | exists %}
+                <Image style="max-width: 90%" src={{src | quote}}
+                    download={{src | replace: ".png", ".pdf" | quote}} />
+            {% else %}
+                <Image style="max-width: 90%" src={{src | quote}} />
+            {% endif %}
         </td>
         {% endif %}
     {% endfor %}

biopipen/scripts/delim/SampleInfo.R

@@ -186,6 +186,15 @@ for (name in names(stats)) {
     print(p)
     dev.off()
 
+    plotfile_pdf <- file.path(outdir, paste0(name, ".pdf"))
+    pdf(
+        plotfile_pdf,
+        width = stat$devpars$width / stat$devpars$res,
+        height = stat$devpars$height / stat$devpars$res
+    )
+    print(p)
+    dev.off()
+
     by_desc <- ifelse(is.null(stat$by), "", paste0(" by ", stat$by))
     descr <- ifelse(
         is_continuous,
@@ -193,7 +202,7 @@ for (name in names(stats)) {
         paste0("The number of ", stat$on, by_desc)
     )
     add_report(
-        list(kind = "table_image", src = plotfile, name = name, descr = descr),
+        list(kind = "table_image", src = plotfile, name = name, descr = descr, download = plotfile_pdf),
         h1 = "Statistics",
         ui = "table_of_images:2"
     )

biopipen/scripts/scrna/CellsDistribution.R

@@ -368,9 +368,17 @@ do_case <- function(name, case) {
     width <- devpars$width %||% (400 + 120 + 100 * ngroups)
     #                         group_by names
     height <- devpars$height %||% (120 + 100 * cells_rows)
+
+    p <- wrap_plots(piecharts, ncol = 1, guides = "collect")
+
     piefile <- file.path(info$casedir, paste0(info$case_slug, ".png"))
     png(piefile, res = res, width = width, height = height)
-    print(wrap_plots(piecharts, ncol = 1, guides = "collect"))
+    print(p)
+    dev.off()
+
+    piefile_pdf <- file.path(info$casedir, paste0(info$case_slug, ".pdf"))
+    pdf(piefile_pdf, width = width / res, height = height / res)
+    print(p)
     dev.off()
 
     log_info("  Plotting and saving heatmap ...")
@@ -411,7 +419,6 @@ do_case <- function(name, case) {
     hm_res <- hm_devpars$res %||% 100
     hm_width <- hm_devpars$width %||% (600 + 15 * length(unique(meta$seurat_clusters)) + extra_width)
     hm_height <- hm_devpars$height %||% (450 + 15 * cells_rows + extra_height)
-    png(hmfile, res = hm_res, width = hm_width, height = hm_height)
     hm <- Heatmap(
         as.matrix(hmdata),
         name = "Size",
@@ -430,6 +437,12 @@ do_case <- function(name, case) {
         right_annotation = row_ha,
         top_annotation = ha
     )
+    png(hmfile, res = hm_res, width = hm_width, height = hm_height)
+    print(hm)
+    dev.off()
+
+    hmfile_pdf <- gsub(".png$", ".pdf", hmfile)
+    pdf(hmfile_pdf, width = hm_width / hm_res, height = hm_height / hm_res)
     print(hm)
     dev.off()
 
@@ -454,11 +467,11 @@ do_case <- function(name, case) {
     add_report(
         list(
             name = "Pie Charts",
-            contents = list(list(kind = "image", src = piefile))
+            contents = list(list(kind = "image", src = piefile, download = piefile_pdf))
         ),
         list(
             name = "Heatmap",
-            contents = list(list(src = hmfile, kind = "image"))
+            contents = list(list(src = hmfile, kind = "image", download = hmfile_pdf))
         ),
         list(
             name = "Distribution Table",
@@ -493,25 +506,37 @@ do_overlap <- function(section) {
     print(venn_p)
     dev.off()
 
+    venn_plot_pdf <- gsub(".png$", ".pdf", venn_plot)
+    pdf(venn_plot_pdf, width = 10, height = 6)
+    print(venn_p)
+    dev.off()
+
     upset_plot <- file.path(sec_dir, "upset.png")
     upset_p <- upset(fromList(overlap_cases))
     png(upset_plot, res = 100, width = 800, height = 600)
     print(upset_p)
     dev.off()
 
+    upset_plot_pdf <- gsub(".png$", ".pdf", upset_plot)
+    pdf(upset_plot_pdf, width = 8, height = 6)
+    print(upset_p)
+    dev.off()
+
     add_report(
         list(
             name = "Venn Plot",
             contents = list(list(
                 kind = "image",
-                src = venn_plot
+                src = venn_plot,
+                download = venn_plot_pdf
             ))
         ),
         list(
             name = "UpSet Plot",
             contents = list(list(
                 kind = "image",
-                src = upset_plot
+                src = upset_plot,
+                download = upset_plot_pdf
             ))
         ),
         h1 = "Overlapping Groups",

biopipen/scripts/scrna/MarkersFinder.R

@@ -208,6 +208,11 @@ plot_volcano = function(markers, volfile, sig, volgenes) {
     png(volfile, res = 100, height = 1200, width = 900)
     print(p_vol)
     dev.off()
+
+    volfile_pdf <- gsub(".png$", ".pdf", volfile)
+    pdf(volfile_pdf, width = 9, height = 12)
+    print(p_vol)
+    dev.off()
 }
 
 # Do enrichment analysis for a case using Enrichr
@@ -256,14 +261,16 @@ do_enrich <- function(info, markers, sig, volgenes) {
                 log_warn("  No enrichment found for case: {info$casename} - {db}")
                 next
             }
-            png(
-                file.path(info$casedir, paste0("Enrichr-", db, ".png")),
-                res = 100, height = 1000, width = 1000
-            )
-            print(
-                plotEnrich(enriched[[db]], showTerms = 20, title = db) +
+            enrich_p <- plotEnrich(enriched[[db]], showTerms = 20, title = db) +
                 theme_prism()
-            )
+            enrich_plot <- file.path(info$casedir, paste0("Enrichr-", db, ".png"))
+            png(enrich_plot, res = 100, height = 1000, width = 1000)
+            print(enrich_p)
+            dev.off()
+
+            enrich_plot_pdf <- gsub(".png$", ".pdf", enrich_plot)
+            pdf(enrich_plot_pdf, width = 10, height = 10)
+            print(enrich_p)
             dev.off()
         }
     }
@@ -287,13 +294,17 @@ do_dotplot <- function(info, siggenes, dotplot, args) {
     dotplot_height <- dotplot_devpars$height %||% 600
     dotplot_res <- dotplot_devpars$res %||% 100
     dotplot_file <- file.path(info$casedir, "dotplot.png")
-    png(dotplot_file, res = dotplot_res, width = dotplot_height, height = dotplot_width)
-    # rotate x axis labels
-    print(
-        do_call(DotPlot, dotplot) +
+    dot_p <- do_call(DotPlot, dotplot) +
         theme(axis.text.x = element_text(angle = 90, hjust = 1)) +
         coord_flip()
-    )
+    png(dotplot_file, res = dotplot_res, width = dotplot_height, height = dotplot_width)
+    # rotate x axis labels
+    print(dot_p)
+    dev.off()
+
+    dotplot_file_pdf <- gsub(".png$", ".pdf", dotplot_file)
+    pdf(dotplot_file_pdf, width = dotplot_height / dotplot_res, height = dotplot_width / dotplot_res)
+    print(dot_p)
     dev.off()
 }
 
@@ -336,9 +347,19 @@ add_case_report <- function(info, sigmarkers, siggenes) {
                 title = "Volcano Plot",
                 ui = "flat",
                 contents = list(
+                    list(
+                        kind = "descr",
+                        content = paste0(
+                            "The volcano plot is generated using ggplot2. ",
+                            "The significant markers are highlighted in blue and red, ",
+                            "indicating up-regulated and down-regulated genes, respectively. ",
+                            "The non-significant genes are shown in grey. "
+                        )
+                    ),
                     list(
                         kind = "img",
-                        src = file.path(info$casedir, "volcano.png")
+                        src = file.path(info$casedir, "volcano.png"),
+                        download = file.path(info$casedir, "volcano.pdf")
                     )
                 )
             ),
@@ -346,9 +367,21 @@ add_case_report <- function(info, sigmarkers, siggenes) {
                 title = "Dot Plot",
                 ui = "flat",
                 contents = list(
+                    list(
+                        kind = "descr",
+                        content = paste0(
+                            "The dot plot is generated using Seurat's DotPlot function. ",
+                            "The top significant markers are used as input. ",
+                            "The sizes of the dots are proportional to the percentage of cells ",
+                            "expressing the gene in each group. ",
+                            "The colors of the dots are proportional to the average expression ",
+                            "of the gene in each group. "
+                        )
+                    ),
                     list(
                         kind = "img",
-                        src = file.path(info$casedir, "dotplot.png")
+                        src = file.path(info$casedir, "dotplot.png"),
+                        download = file.path(info$casedir, "dotplot.pdf")
                     )
                 )
             ),
@@ -613,6 +646,14 @@ do_overlap <- function(section) {
         do.call(png, ov_args$venn$devpars)
         print(venn_p)
         dev.off()
+
+        venn_plot_pdf <- file.path(ov_dir, "venn.pdf")
+        pdf(
+            venn_plot_pdf,
+            width = ov_args$venn$devpars$width / ov_args$venn$devpars$res,
+            height = ov_args$venn$devpars$height / ov_args$venn$devpars$res)
+        print(venn_p)
+        dev.off()
     }
 
     df_markers <- fromList(ov_cases)
@@ -654,6 +695,14 @@ do_overlap <- function(section) {
         do.call(png, ov_args$upset$devpars)
         print(upset_p)
         dev.off()
+
+        upset_plot_pdf <- file.path(ov_dir, "upset.pdf")
+        pdf(
+            upset_plot_pdf,
+            width = ov_args$upset$devpars$width / ov_args$upset$devpars$res,
+            height = ov_args$upset$devpars$height / ov_args$upset$devpars$res)
+        print(upset_p)
+        dev.off()
     }
 
     add_report(
@@ -663,7 +712,8 @@ do_overlap <- function(section) {
             contents = list(
                 list(
                     kind = "img",
-                    src = file.path(ov_dir, "venn.png")
+                    src = file.path(ov_dir, "venn.png"),
+                    download = file.path(ov_dir, "venn.pdf")
                 )
             )
         ),
@@ -673,7 +723,8 @@ do_overlap <- function(section) {
             contents = list(
                 list(
                     kind = "img",
-                    src = file.path(ov_dir, "upset.png")
+                    src = file.path(ov_dir, "upset.png"),
+                    download = file.path(ov_dir, "upset.pdf")
                 )
             )
         ),

biopipen/scripts/scrna/MetaMarkers.R

@@ -171,14 +171,17 @@ do_enrich <- function(info, markers, sig) {
             next
         }
 
-        png(
-            file.path(info$casedir, paste0("Enrichr-", db, ".png")),
-            res = 100, height = 600, width = 800
-        )
-        print(
-            plotEnrich(enriched[[db]], showTerms = 20, title = db) +
+        p <- plotEnrich(enriched[[db]], showTerms = 20, title = db) +
             theme_prism()
-        )
+
+        plotfile <- file.path(info$casedir, paste0("Enrichr-", db, ".png"))
+        png(plotfile, res = 100, height = 600, width = 800)
+        print(p)
+        dev.off()
+
+        plotfile_pdf <- gsub(".png$", ".pdf", plotfile)
+        pdf(plotfile_pdf, height = 6, width = 8)
+        print(p)
         dev.off()
     }
 }
@@ -292,9 +295,15 @@ do_case <- function(casename) {
             print(p)
             dev.off()
 
+            outfile_pdf <- gsub(".png$", ".pdf", outfile)
+            pdf(outfile_pdf, height = 6, width = 8)
+            print(p)
+            dev.off()
+
             geneplots[[length(geneplots) + 1]] <- list(
                 kind = "table_image",
                 src = outfile,
+                download = outfile_pdf,
                 name = gene
             )
         }

biopipen/scripts/scrna/RadarPlots.R

@@ -193,6 +193,7 @@ do_radarplot <- function(info, case, counts) {
 
     # Plot
     plotfile = file.path(info$casedir, "plot.png")
+    plotfile_pdf = file.path(info$casedir, "plot.pdf")
     if (!is.null(case$colors) && length(case$colors) == 1 && case$colors == "biopipen") {
         colors = pal_biopipen()(nrow(rdr_data))
     } else if (!is.null(case$colors) && length(case$colors) > 0) {
@@ -217,6 +218,14 @@ do_radarplot <- function(info, case, counts) {
     )
     print(p)
     dev.off()
+
+    pdf(
+        plotfile_pdf,
+        width = case$devpars$width / case$devpars$res,
+        height = case$devpars$height / case$devpars$res
+    )
+    print(p)
+    dev.off()
 }
 
 do_barplot_and_tests <- function(info, case, counts) {
@@ -268,6 +277,7 @@ do_barplot_and_tests <- function(info, case, counts) {
 
     # Plot the barplot
     plotfile = file.path(info$casedir, "barplot.png")
+    plotfile_pdf = file.path(info$casedir, "barplot.pdf")
     p = ggplot(plotdata, aes(x = !!sym(case$ident), y = .mean, fill = !!sym(case$by))) +
         geom_bar(stat = "identity", position = "dodge") +
         geom_errorbar(
@@ -295,6 +305,14 @@ do_barplot_and_tests <- function(info, case, counts) {
     print(p)
     dev.off()
 
+    pdf(
+        plotfile_pdf,
+        width = case$bar_devpars$width / case$bar_devpars$res,
+        height = case$bar_devpars$height / case$bar_devpars$res
+    )
+    print(p)
+    dev.off()
+
     # Do the tests in each cluster between groups on .frac
     bys <- bardata %>% pull(!!sym(case$by)) %>% unique()
     if (!is.null(case$test) && test != "none") {
@@ -348,7 +366,8 @@ add_case_report = function(info, breakdown, test) {
             contents = list(
                 list(
                     kind = "image",
-                    src = file.path(info$casedir, "plot.png")
+                    src = file.path(info$casedir, "plot.png"),
+                    download = file.path(info$casedir, "plot.pdf")
                 )
             )
         ),
@@ -381,7 +400,8 @@ add_case_report = function(info, breakdown, test) {
                 contents = list(
                     list(
                         kind = "image",
-                        src = file.path(info$casedir, "barplot.png")
+                        src = file.path(info$casedir, "barplot.png"),
+                        download = file.path(info$casedir, "barplot.pdf")
                     )
                 )
             ))

biopipen/scripts/scrna/SeuratClusterStats-clustree.R

@@ -46,9 +46,14 @@ if (
             command <- srtobj@commands[[paste0("FindClusters.", prefix)]] %||%
                 (if(prefix == "seurat_clusters") srtobj@commands$FindClusters else NULL)
 
+            p <- do_call(clustree, case)
             clustree_file <- file.path(odir, paste0(prefix, ".clustree.png"))
             png(clustree_file, width = devpars$width, height = devpars$height, res = devpars$res)
-            p <- do_call(clustree, case)
+            print(p)
+            dev.off()
+
+            clustree_file_pdf <- file.path(odir, paste0(prefix, ".clustree.pdf"))
+            pdf(clustree_file_pdf, width = devpars$width / devpars$res, height = devpars$height / devpars$res)
             print(p)
             dev.off()
 
@@ -62,6 +67,7 @@ if (
             reports[[length(reports) + 1]] <- list(
                 kind = "table_image",
                 src = clustree_file,
+                download = clustree_file_pdf,
                 name = name,
                 descr = paste0("Resolutions: ", paste(resolution, collapse = ", "), "; resolution used: ", resolution_used)
             )

biopipen/scripts/scrna/SeuratClusterStats-dimplots.R

@@ -49,9 +49,15 @@ do_one_dimplot = function(name) {
 
     if (case$reduction %in% c("dim", "auto")) { case$reduction = NULL }
     case$na_group <- NULL
+
+    p = do_call(DimPlot, case)
     figfile = file.path(odir, paste0(slugify(name), ".dim.png"))
     png(figfile, width=devpars$width, height=devpars$height, res=devpars$res)
-    p = do_call(DimPlot, case)
+    print(p)
+    dev.off()
+
+    figfile_pdf = file.path(odir, paste0(slugify(name), ".dim.pdf"))
+    pdf(figfile_pdf, width=devpars$width / devpars$res, height=devpars$height / devpars$res)
     print(p)
     dev.off()
 
@@ -62,7 +68,8 @@ do_one_dimplot = function(name) {
         ),
         list(
             kind = "image",
-            src = figfile
+            src = figfile,
+            download = figfile_pdf
         ),
         h1 = name
     )

biopipen/scripts/scrna/SeuratClusterStats-features.R

@@ -224,6 +224,7 @@ do_one_features = function(name) {
 
     if (kind == "bar") {
         figfile <- file.path(odir, paste0(slugify(name), ".bar.png"))
+        figfile_pdf <- file.path(odir, paste0(slugify(name), ".bar.pdf"))
         genes <- rownames(GetAssayData(case$object))
         genes <- genes[sapply(genes, function(x) grepl(x, case$features))]
         if (length(genes) == 0) {
@@ -265,6 +266,10 @@ do_one_features = function(name) {
         print(p)
         dev.off()
 
+        pdf(figfile_pdf, width = devpars$width / devpars$res, height = devpars$height / devpars$res)
+        print(p)
+        dev.off()
+
         add_report(
             list(
                 kind = "descr",
@@ -272,7 +277,8 @@ do_one_features = function(name) {
             ),
             list(
                 kind = "image",
-                src = figfile
+                src = figfile,
+                download = figfile_pdf
             ),
             h1 = ifelse(is.null(section), name, section),
             h2 = ifelse(is.null(section), "#", name)
@@ -284,6 +290,7 @@ do_one_features = function(name) {
     case$features = .get_features(case$features)
     if (kind == "avgheatmap") {
         figfile <- file.path(odir, paste0(slugify(name), ".avgheatmap.png"))
+        figfile_pdf <- file.path(odir, paste0(slugify(name), ".avgheatmap.pdf"))
         assay <- assay %||% DefaultAssay(object)
         layer <- layer %||% ifelse("scale.data" %in% Layers(object, assay = assay), "scale.data", "data")
 
@@ -353,6 +360,10 @@ do_one_features = function(name) {
         print(p)
         dev.off()
 
+        pdf(figfile_pdf, width = devpars$width / devpars$res, height = devpars$height / devpars$res)
+        print(p)
+        dev.off()
+
         add_report(
             list(
                 kind = "descr",
@@ -360,7 +371,8 @@ do_one_features = function(name) {
             ),
             list(
                 kind = "image",
-                src = figfile
+                src = figfile,
+                download = figfile_pdf
             ),
             h1 = ifelse(is.null(section), name, section),
             h2 = ifelse(is.null(section), "#", name)
@@ -430,6 +442,11 @@ do_one_features = function(name) {
     })
     dev.off()
 
+    figfile_pdf = gsub(".png$", ".pdf", figfile)
+    pdf(figfile_pdf, width=devpars$width / devpars$res, height=devpars$height / devpars$res)
+    print(p)
+    dev.off()
+
     add_report(
         list(
             kind = "descr",
@@ -437,7 +454,8 @@ do_one_features = function(name) {
         ),
         list(
             kind = "image",
-            src = figfile
+            src = figfile,
+            download = figfile_pdf
         ),
         h1 = ifelse(is.null(section), name, section),
         h2 = ifelse(is.null(section), "#", name)

biopipen/scripts/scrna/SeuratClusterStats-hists.R

@@ -6,6 +6,7 @@ log_info("hists:")
 
 do_one_hists <- function(m, case, odir, h1, each = NULL) {
     ofile <- file.path(odir, paste0(slugify(h1), ifelse(is.null(each), "", paste0("-", slugify(each))), ".png"))
+    ofile_pdf <- gsub(".png$", ".pdf", ofile)
 
     p <- ggplot(m, aes(x=!!sym(case$x))) +
         geom_histogram(
@@ -42,16 +43,20 @@ do_one_hists <- function(m, case, odir, h1, each = NULL) {
     print(p)
     dev.off()
 
+    pdf(ofile_pdf, width=devpars$width / devpars$res, height=devpars$height / devpars$res)
+    print(p)
+    dev.off()
+
     # Add report
     if (!is.null(each)) {
         add_report(
-            list(src = ofile, descr = first(m[[case$each]])),
+            list(src = ofile, descr = first(m[[case$each]]), download = ofile_pdf),
             h1 = h1,
             ui = "table_of_images"
         )
     } else {
         add_report(
-            list(kind = "image", src = ofile),
+            list(kind = "image", src = ofile, download = ofile_pdf),
             h1 = h1
         )
     }