@@ -80,7 +80,7 @@ Imports:

     Seurat (>= 3.1.3),

     shiny,

     shinyjs,

-    SingleR,

+    SingleR (>= 1.99.2),

     SoupX,

     sva,

     reshape2,

@@ -43,10 +43,10 @@ runSingleR <- function(inSCE,

                        featureType = c("symbol", "ensembl"),

                        labelByCluster = NULL) {

     # Input checks

-    if(!inherits(inSCE, "SingleCellExperiment")){

+    if (!inherits(inSCE, "SingleCellExperiment")) {

         stop('"inSCE" should be a SingleCellExperiment inherited Object.')

     }

-    if(!useAssay %in% expDataNames(inSCE)){

+    if (!useAssay %in% expDataNames(inSCE)) {

         stop('"useAssay" name: ', useAssay, ' not found.')

     }

@@ -91,7 +91,7 @@ runSingleR <- function(inSCE,

             if (!isTRUE(useEnsembl)) {

                 rownames(ref) <- SummarizedExperiment::rowData(ref)$symbol

             }

-            ref <- ref[,!is.na(ref$label) & ref$label!="unclear"]

+            ref <- ref[,!is.na(ref$label) & ref$label != "unclear"]

             ref <- scaterlogNormCounts(ref, assayName = "logcounts")

             labelColName <- "label"

             warning("MuraroPancreasData does not have multiple levels of ",

@@ -375,20 +375,21 @@ Ontology information. For more information, please click on the function

 name and see the reference page of this function.</p>

 <div class="sourceCode" id="cb2"><pre class="downlit sourceCode r">

 <code class="sourceCode R"><span><span class="va">sce</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/runSingleR.html">runSingleR</a></span><span class="op">(</span>inSCE <span class="op">=</span> <span class="va">sce</span>, useAssay <span class="op">=</span> <span class="st">"logcounts"</span>, useBltinRef <span class="op">=</span> <span class="st">"hpca"</span>, level <span class="op">=</span> <span class="st">"main"</span>, featureType <span class="op">=</span> <span class="st">"symbol"</span><span class="op">)</span></span></code></pre></div>

-<p>Four results from SingleR will be stored in <code><a href="https://rdrr.io/pkg/SummarizedExperiment/man/SummarizedExperiment-class.html" class="external-link">colData()</a></code>

+<p>Four results from SingleR will be stored in <code>colData()</code>

 slot with the same prefix as

 <code>"SingleR_{reference abbr}_{annotation level}_"</code>, then

-<code>"score"</code>, <code>"first.labels"</code>, <code>"labels"</code>

-and <code>"pruned.labels"</code>, respectively.

-<code>"first.labels"</code> refers to the labeling initially indicated

-by the scores, <code>"labels"</code> is fine-tuned, and

-<code>"pruned.labels"</code> is the pruned result.</p>

+<code>"score"</code>, <code>"labels"</code>, <code>"delta.next"</code>

+and <code>"pruned.labels"</code>, respectively. <code>"labels"</code>

+contains the predicted label, basing only on the maximum entry in

+scores. <code>"delta.next"</code> contains difference between the best

+and next-best score. <code>"pruned.labels"</code> contains predictions

+where “low-quality” labels are replaced with <code>NA</code>s.</p>

 <p>Besides, users can also use their own labeled reference datasets

 wrapped in a <a href="https://rdrr.io/bioc/SingleCellExperiment/man/SingleCellExperiment.html" class="external-link">SingleCellExperiment</a>

 object. Refer to argument <code>useSCERef</code> and

 <code>labelColName</code>. Additionally, the labeling can also be done

 on cluster label if users have already performed clustering on their

-dataset and have the result stored in <code><a href="https://rdrr.io/pkg/SummarizedExperiment/man/SummarizedExperiment-class.html" class="external-link">colData()</a></code> slot. Refer

+dataset and have the result stored in <code>colData()</code> slot. Refer

 to argument <code>labelByCluster</code>.</p>

 <p><strong>Visualization</strong></p>

 <p>Users can choose to visualize the labeling result on a scatter plot.

Binary files a/docs/articles/cell_type_labeling_files/figure-html/visualization-1.png and b/docs/articles/cell_type_labeling_files/figure-html/visualization-1.png differ

@@ -7,10 +7,17 @@ sce <- scaterlogNormCounts(sce, "logcounts")

 rownames(sce) <- rowData(sce)$feature_name

 test_that(desc = "Testing SingleR", {

-  sce <- runSingleR(sce)

-

-  testthat::expect_true("SingleR_hpca_main_pruned.labels" %in% names(colData(sce)))

-  testthat::expect_true("SingleR_hpca_main_labels" %in% names(colData(sce)))

-  testthat::expect_true("SingleR_hpca_main_first.labels" %in% names(colData(sce)))

-  testthat::expect_true("SingleR_hpca_main_scores" %in% names(colData(sce)))

+  tryCatch({

+    ref <- celldex::HumanPrimaryCellAtlasData()

+  }, error = function(e) {

+    message("Error importing reference with `celldex` library. ",

+            "Skipping runSingleR test.")

+  }, finally = {

+    sce <- runSingleR(sce)

+    testthat::expect_true("SingleR_hpca_main_scores" %in% names(colData(sce)))

+    testthat::expect_true("SingleR_hpca_main_labels" %in% names(colData(sce)))

+    #testthat::expect_true("SingleR_hpca_main_first.labels" %in% names(colData(sce)))

+    #testthat::expect_true("SingleR_hpca_main_delta.next" %in% names(colData(sce)))

+    testthat::expect_true("SingleR_hpca_main_pruned.labels" %in% names(colData(sce)))

+  })

 })

@@ -94,7 +94,7 @@ Then users can use `runSingleR()` to invoke the algorithm. Here we choose to use

 sce <- runSingleR(inSCE = sce, useAssay = "logcounts", useBltinRef = "hpca", level = "main", featureType = "symbol")

 ```

-Four results from SingleR will be stored in `colData()` slot with the same prefix as `"SingleR_{reference abbr}_{annotation level}_"`, then `"score"`, `"first.labels"`, `"labels"` and `"pruned.labels"`, respectively. `"first.labels"` refers to the labeling initially indicated by the scores, `"labels"` is fine-tuned, and `"pruned.labels"` is the pruned result.  

+Four results from SingleR will be stored in `colData()` slot with the same prefix as `"SingleR_{reference abbr}_{annotation level}_"`, then `"score"`, `"labels"`, `"delta.next"` and `"pruned.labels"`, respectively. `"labels"` contains the predicted label, basing only on the maximum entry in scores. `"delta.next"` contains difference between the best and next-best score. `"pruned.labels"` contains predictions where "low-quality" labels are replaced with `NA`s. 

 Besides, users can also use their own labeled reference datasets wrapped in a [SingleCellExperiment](https://rdrr.io/bioc/SingleCellExperiment/man/SingleCellExperiment.html) object. Refer to argument `useSCERef` and `labelColName`. Additionally, the labeling can also be done on cluster label if users have already performed clustering on their dataset and have the result stored in `colData()` slot. Refer to argument `labelByCluster`.