Feat/pathfinder

NAMESPACE

@@ -38,6 +38,8 @@ export(convert_to_stringlist)
 export(copy_annotations)
 export(data_curator_app_subpage)
 export(delete_provenance)
+export(find_child)
+export(find_in)
 export(from_pubmed)
 export(get_by_prop_from_json_schema)
 export(get_dependency_from_json_schema)
@@ -57,8 +59,10 @@ export(missing_annotation_email)
 export(new_dataset)
 export(new_project)
 export(nf_cnv_dataset)
+export(nf_find_asset)
 export(nf_sarek_datasets)
 export(nf_star_salmon_datasets)
+export(nf_workflow_version)
 export(processing_flowchart)
 export(qc_manifest)
 export(register_study)

R/find.R

@@ -0,0 +1,90 @@
+# Utils to help overcome nested folders
+
+#' Find in path
+#' 
+#' Get the Synapse id of an entity nested several folder layers deep without
+#' having to click through the UI or create a fileview as long as the structure/path is known.
+#' 
+#' @param scope Id of the container (project or folder) to begin search.
+#' @param path Path string in format "subdir1/subdir2/file.txt", where the last-level element will be the id returned.
+#' @export
+find_in <- function(scope, path) {
+
+  path_list <- strsplit(path, split = "/", fixed = TRUE)[[1]]
+  here <- scope
+  id <- NULL
+  while(length(path_list)) {
+    child <- first(path_list)
+    path_list <- path_list[-1]
+    here <- find_child(child_name = child, parent = here)
+    id <- here
+  }
+  id
+}
+
+#' Find id of a child entity in a container
+#'
+#' @param parent Parent container (project or folder).
+#' @param child_name Name of child entity.
+#' @export
+find_child <- function(child_name, parent) {
+
+  q <- .syn$getChildren(parent)
+  child_id <- NULL
+  repeat {
+    x <- reticulate::iter_next(q) 
+    if(is.null(x) || x$name == child_name) {
+      child_id <- x$id
+      break
+    }
+  }
+  child_id
+}
+
+# Find nextflow assets --------------------------------------------------------- #
+
+# Convenience functions for getting Synapse ids of nextflow assets
+
+#' Find a standard nextflow workflow output asset
+#' 
+#' Note that samplesheets became part of the output only for newer versions of nf-core/rna-seq;
+#' older runs may not find samplesheets. 
+#' Paths default to known working paths corresponding to the latest major workflow version, 
+#' but this may change and may need to be updated as part of util maintenance. 
+#' 
+#' @param syn_out Id of top-level folder that corresponds to `publishDir` in a nextflow workflow.
+#' @param asset Name of asset to find.
+#' @returns Id of samplesheet. 
+#' @export
+nf_find_asset <- function(syn_out, 
+                          asset = c("software_versions", "multiqc_report", "samplesheet", "samtools_stats")) {
+
+  asset <- match.arg(asset)
+  path <- switch(asset,
+                 software_versions = "pipeline_info/software_versions.yml",
+                 multiqc_report = "multiqc/star_salmon/multiqc_report.html",
+                 samplesheet = "pipeline_info/samplesheet.valid.csv",
+                 samtools_stats = "multiqc/star_salmon/multiqc_data/multiqc_samtools_stats.txt"
+  )
+
+  id <- find_in(syn_out, path)
+  if(is.null(id)) stop("File not found. Is this the right output directory/path?")
+  id
+}
+
+
+#' Return workflow version according to workflow meta
+#' 
+#' @inheritParams nf_find_asset
+#' @returns Version string. 
+#' @export
+nf_workflow_version <- function(syn_out) {
+
+  version_meta <- nf_find_asset(syn_out, asset = "software_versions")
+  file <- .syn$get(version_meta, downloadFile = TRUE)
+  yml <- yaml::read_yaml(file$path)
+  workflow <- grep("nf-core", names(yml$Workflow))
+  yaml$Workflow[[workflow]]
+
+}
+

_pkgdown.yml

@@ -44,6 +44,7 @@ reference:
   - annotate_cnv
   - annotate_reports_sarek
   - annotate_with_tool_stats
+  - nf_workflow_version
 
 - title: Dataset Creation and Management
 - subtitle: General dataset creation and citation
@@ -80,6 +81,12 @@ reference:
   - summarize_file_access
   - grant_specific_file_access
 
+- title: Search Utils
+  desc: Help locate Synapse accessions, etc.
+- contents: 
+  - contains("find")
+  - nf_find_asset
+
 - title: Provenance Utils
   desc: Manage provenance metadata
 - contents: