Merge pull request #29 from mc2-center/allow-custom-config

Allow use of a custom config file

README.md

@@ -61,9 +61,11 @@ Other columns may be provided but are not used by the pipeline.
 - `input`: Path to a CSV sample sheet. This parameter is required.
 - `outDir`: Specifies the directory where the output data should be saved. Default is `outputs`.
 
-#### HistoQC options
+#### Other options
 
-- `config`: Configuration file used by HistoQC. Must be one of `default`, `ihc`, `clinical`, `first`, `light`, or `v2.1`.Custom config files are not currently supported.
+- `config` (string): Name of a built-in configuration used by HistoQC. Must be one of `default`, `ihc`, `clinical`, `first`, `light`, or `v2.1`. Defaults to `default`.
+- `custom_config` (path): Path to a HistoQC compatible configuration file. Must have a `.ini` extension. Overrides `config`.
+- `convert` (bool): If provided, `vips` is used to [create an OpenSlide compatiable TIFF file](http://www.andrewjanowczyk.com/converting-an-existing-image-into-an-openslide-compatible-format/). Uses [mc2-center/histoqc-openslide-converter](https://github.com/mc2-center/histoqc-openslide-converter).
 
 ### Profiles
 
@@ -86,23 +88,45 @@ The container is automatically pulled by NextFlow, but if local use is required
 > A Nextflow pipeline is implicitly modelled by a direct acyclic graph (DAG). The vertices in the graph represent the pipeline’s processes and operators, while the edges represent the data connections (i.e. channels) between them.
 
 ```mermaid
-flowchart TD
-    p0((Channel.fromPath))
-    p1([splitCsv])
-    p2([map])
-    p3[NF_HISTOQC:RUN_HISTOQC:HISTOQC]
-    p4(( ))
-    p5(( ))
-    p6([collect])
-    p7[NF_HISTOQC:COLLECT_RESULTS:COLLECT]
-    p8(( ))
-    p0 --> p1
-    p1 --> p2
-    p2 -->|images| p3
-    p3 -->|results| p6
-    p3 --> p5
-    p3 -->|output| p4
-    p6 --> p7
-    p7 --> p8
+flowchart TB
+    subgraph " "
+    v0["Channel.fromPath"]
+    v3["Channel.fromPath"]
+    v6["config_string"]
+    end
+    subgraph NF_HISTOQC
+    subgraph RUN
+    v5([CONVERT])
+    v7([HISTOQC])
+    v1(( ))
+    v4(( ))
+    v9(( ))
+    v13(( ))
+    end
+    subgraph COLLECT
+    v10([RESULTS])
+    v11([TIDY])
+    v14([LOGS])
+    end
+    end
+    subgraph " "
+    v8["output"]
+    v12[" "]
+    v15[" "]
+    end
+    v0 --> v1
+    v3 --> v4
+    v1 --> v5
+    v5 --> v7
+    v6 --> v7
+    v4 --> v7
+    v7 --> v8
+    v7 --> v9
+    v7 --> v13
+    v9 --> v10
+    v10 --> v11
+    v11 --> v12
+    v13 --> v14
+    v14 --> v15
 
 ```

main.nf

@@ -2,24 +2,35 @@
 
 nextflow.enable.dsl=2
 
+// Set parameters and defaults
+
+// Check if the input samplesheet is provided
 if (params.input) { params.input = file(params.input) } else { exit 1, 'Input samplesheet not specified!' }
 
+// Check requirements for the convert param to be used
 params.convert = false
-
 if (params.convert) {
     if (!params.mag || !params.mpp) {
         error "Both nominal magnification ('mag') and microns per pixel ('mpp' parameters must be provided when 'params.convert' is set to true."
     }
 }
 
+// Check requirements for the custom_config param to be used
+params.custom_config = null
+if ( params.custom_config ) {
+    if (! file(params.custom_config).exists() ) { error "Custom config file does not exist" }
+    if (! file(params.custom_config).getName().endsWith(".ini") ) { error "Custom config file is not a .ini file" }
+}
 
-// Set parameters and defaults
+// Set default outdir and config
 params.outDir = './outputs'
 params.config = 'default'
 
+// Import workflow
 include { NF_HISTOQC } from './workflows/nf_histoqc.nf'
-// include { COMBINE_RESULTS } from './modules/combine_results.nf'
 
+
+// Run workflow
 workflow {
     NF_HISTOQC ()
 }

modules/histoqc.nf

@@ -8,6 +8,8 @@ process HISTOQC {
 
     input:
     tuple val(meta), path(images)
+    val config_string
+    path custom_config
 
     output:
     path "out/results.tsv", emit: results
@@ -16,7 +18,18 @@ process HISTOQC {
 
     script:
     """
-    python -m histoqc -c ${params.config} $images -o out
+    # Add the logic for the if config_string or custom_config here
+    # Check if custom_config points to "empty.txt"
+    if [[ "$custom_config" == *empty.txt ]]; then
+        # If yes, use the config_string
+        ini="$config_string"
+    else
+        # If no, use the path to custom_config
+        ini=$custom_config
+    fi
+
+    echo "Using config: \$ini"
+    python -m histoqc $images -o out -c \$ini
     mv out/$images/*.png .
     """
 }

nextflow_schema.json

@@ -29,13 +29,20 @@
       "title": "HistoQC options",
       "type": "object",
       "fa_icon": "fas fa-solid fa-microscope",
-      "description": "Define parameters for histoqc",
+      "description": "Define parameters for HistoQC",
       "required": ["config"],
       "properties": {
         "config": {
           "type": "string",
           "default": "default",
-          "enum": ["default", "ihc", "clinical", "first", "light", "v2.1"]
+          "enum": ["default", "ihc", "clinical", "first", "light", "v2.1"],
+          "description": "Name of built-in HistoQC config. Overridden by `custom_config`."
+        },
+        "custom_config": {
+          "type": "string",
+          "format": "file-path",
+          "mimetype": "text/plain",
+          "description": "Path to a HistoQC config `.ini` file. Overrides `config`"
         }
       }
     },

subworkflows/run_histoqc.nf

@@ -8,7 +8,18 @@ workflow RUN {
 
     main:
 
-    params.convert ? CONVERT(images) | HISTOQC : HISTOQC(images)
+    // If a custom config is passed use it, otherwise pass a dummy file
+    def custom_config = params.custom_config ? Channel.fromPath(params.custom_config).first() : Channel.fromPath("assets/empty.txt").first() // need to pass a dummy file in the projectDir or similar
+
+    // Scenario 1: Conversion
+    if (params.convert ) {
+        CONVERT(images)
+        HISTOQC(CONVERT.out, params.config, custom_config)
+    } 
+    // Scenario 2: No conversion
+    else {
+        HISTOQC(images, params.config, custom_config)
+    }
 
     emit:
     output = HISTOQC.out.masks