Merge pull request #2 from milaboratory/calculate-pfconv-params

calculate-pfconv-params
platforma-open · Sep 7, 2024 · 945be49 · 945be49
2 parents be2ffab + 9144a01
commit 945be49
Show file tree

Hide file tree

Showing 17 changed files with 1,596 additions and 171 deletions.
diff --git a/test/src/wf.test.ts b/test/src/wf.test.ts
@@ -166,7 +166,7 @@ blockTest(
     const clonesPfColumnList = await ml.driverKit.pFrameDriver.listColumns(clonesPfHandle);
 
     // console.log(clonesPfColumnList);
-    expect(clonesPfColumnList).length.to.greaterThanOrEqual(1);
+    expect(clonesPfColumnList).length.to.greaterThanOrEqual(7);
 
     // console.dir(clonotypingStableState3, { depth: 8 });
   }

diff --git a/test/test_config.json b/test/test_config.json
diff --git a/workflow/index.d.ts b/workflow/index.d.ts
@@ -1,4 +1,4 @@
 declare type TemplateFromFile = { readonly type: "from-file"; readonly path: string; };
-declare type TplName = "calculate-pfconv-params" | "export-clones" | "get-preset-for-back" | "get-preset" | "list-presets" | "mixcr-analyze" | "prerun" | "process" | "test.columns.test" | "main";
+declare type TplName = "calculate-pfconv-params" | "export-clones" | "get-preset" | "list-presets" | "mixcr-analyze" | "prerun" | "process" | "test.columns.test" | "main";
 declare const Templates: Record<TplName, TemplateFromFile>;
 export { Templates };
diff --git a/workflow/index.js b/workflow/index.js
@@ -1,7 +1,6 @@
 module.exports = { Templates: {
   'calculate-pfconv-params': { type: 'from-file', path: require.resolve('./dist/tengo/tpl/calculate-pfconv-params.plj.gz') },
   'export-clones': { type: 'from-file', path: require.resolve('./dist/tengo/tpl/export-clones.plj.gz') },
-  'get-preset-for-back': { type: 'from-file', path: require.resolve('./dist/tengo/tpl/get-preset-for-back.plj.gz') },
   'get-preset': { type: 'from-file', path: require.resolve('./dist/tengo/tpl/get-preset.plj.gz') },
   'list-presets': { type: 'from-file', path: require.resolve('./dist/tengo/tpl/list-presets.plj.gz') },
   'mixcr-analyze': { type: 'from-file', path: require.resolve('./dist/tengo/tpl/mixcr-analyze.plj.gz') },

diff --git a/workflow/src/calculate-pfconv-params.tpl.tengo b/workflow/src/calculate-pfconv-params.tpl.tengo
@@ -1,36 +1,87 @@
 // calculate-pfconv-params
-
 self := import("@milaboratory/tengo-sdk:tpl")
 
 pvconfParams := import(":pfconv_params")
 
+json := import("json")
+
 self.defineOutputs("params")
 
 self.body(func(inputs) {
-	// preset content
-	preset := inputs.preset
+
+	// filter function
+	filter := func(list, predicate) {
+    	result := []
+    	for item in list {
+        	if predicate(item) {
+            	result = append(result, item)
+        	}
+    	}
+    	return result
+	}
+
+	fieldsToColumns := {
+    	"-cellGroup": "cellGroup",
+		"-topChains": "topChains",
+    	"-isotype": "isotype",
+    	"-vHit": "bestVHit",
+    	"-dHit": "bestDHit",
+    	"-jHit": "bestJHit",
+    	"-cHit": "bestCHit"
+		}
+
+	// get specs and preset data
+	specs := inputs.specs.getDataAsJson()
+	preset := inputs.preset.getDataAsJson()
+	//if using readCount or umiCount
+	columnsList := []
+	columnsOrder := []
+	if len(specs.umiTags) == 0 {
+		columnsList += ["readCount", "readFraction"]
+	} else {
+		columnsList += ["uniqueMoleculeCount", "uniqueMoleculeFraction"]
+	}
+
+	// check if imputed or not and filter columns by fieldsToColumns
+	imputed := undefined
+	for field in preset.exportClones.fields {
+		if field.field == "-allAAFeaturesImputed" {
+			imputed = true
+		} else if field.field == "-allAAFeatures" {
+			imputed = false
+		} else if !is_undefined(fieldsToColumns[field.field]) {
+			columnsList = append(columnsList, fieldsToColumns[field.field])
+		}			
+	}
+
+	// add columns with sequences
+	for feature in specs.coveredFeaturesOnExport {
+		if imputed {
+			columnsList += ["nSeqImputed"+feature, "minQual"+feature, "aaSeqImputed"+feature]
+		} else {
+			columnsList += ["nSeq"+feature, "minQual"+feature, "aaSeq"+feature]
+		}
+	}
+
+	// filter pvconfParams by column name and get a list of fields from it
+	columns := []
+	for col in columnsList {
+		getColumn := filter(pvconfParams.columns, func(item) { return item.column ==  col})
+		columns += getColumn
+	}
+
+	// create axes fields
+	axes := []
+	if len(specs.cellTags) == 0 {
+		axes = filter(pvconfParams.axes, func(item) { return item.column == "cloneId" }) 
+	} else {
+		axes = pvconfParams.axes
+	}
 
 	return {
 		params: {
-			"axes": [{
-				"column": "cloneId",
-				"spec": {
-					"name": "pl7.app/vdj/cloneId",
-					"type": "Long"
-				}
-			}],
-			"columns": [{
-				"column": "readCount",
-				"id": "read-count",
-				"allowNA": false,
-				"spec": {
-					"name": "pl7.app/vdj/readCount",
-					"valueType": "Long",
-					"annotations": {
-						"pl7.app/label": "Number Of Reads"
-					}
-				}
-			}],
+			"axes": axes,
+			"columns": columns,
 			"storageFormat": "Binary",
   			"partitionKeyLength": 0
 		}

diff --git a/workflow/src/get-preset-for-back.tpl.tengo b/workflow/src/get-preset-for-back.tpl.tengo
diff --git a/workflow/src/get-preset.tpl.tengo b/workflow/src/get-preset.tpl.tengo
@@ -4,6 +4,7 @@ self := import("@milaboratory/tengo-sdk:tpl")
 ll := import("@milaboratory/tengo-sdk:ll")
 exec := import("@milaboratory/tengo-sdk:exec")
 assets := import("@milaboratory/tengo-sdk:assets")
+mixcr := import(":mixcr")
 
 self.validateInputs({
 	"__options__,closed": "",
@@ -16,30 +17,46 @@ self.validateInputs({
 
 mixcrSw := assets.importSoftware("@milaboratory/mixcr:main")
 
-self.defineOutputs("preset")
+self.defineOutputs("preset", "presetSpecForBack")
 
 self.body(func(inputs) {
 	presetName := inputs.request.presetName
 	species := inputs.request.species
 
-	mixcrCmdBuilder := exec.builder().
+	mixcrExportPresetCmdBuilder := exec.builder().
 						software(mixcrSw).
 						printErrStreamToStdout().
 						arg("exportPreset").
 						arg("--preset-name").
 						arg(presetName)
 
+	mixcr.addAdditionalColumnArgs(mixcrExportPresetCmdBuilder)
+
 	if !is_undefined(species) {
-		mixcrCmdBuilder.arg("--species").arg(species)
+		mixcrExportPresetCmdBuilder.arg("--species").arg(species)
 	}
 
-	mixcrCmd := mixcrCmdBuilder.arg("preset.json").
+	mixcrExportPresetCmd := mixcrExportPresetCmdBuilder.arg("preset.json").
 						saveFileContent("preset.json").
+						saveFile("preset.json").
+						run()
+
+	presetContent := mixcrExportPresetCmd.getFileContent("preset.json")
+
+	mixcrForBackCmd := exec.builder().
+						software(mixcrSw).
+						printErrStreamToStdout().
+						arg("presetSpecificationsForBack").
+						addFile("preset.yaml", mixcrExportPresetCmd.getFile("preset.json")).
+						arg("preset.yaml").
+						arg("presetForBack.json").
+						saveFileContent("presetForBack.json").
 						run()
 
-	presetContent := mixcrCmd.getFileContent("preset.json")
+	presetSpecForBackContent := mixcrForBackCmd.getFileContent("presetForBack.json")
 
 	return {
-		preset: presetContent
+		preset: presetContent,
+		presetSpecForBack: presetSpecForBackContent
 	}
 })
diff --git a/workflow/src/main.tpl.tengo b/workflow/src/main.tpl.tengo
@@ -13,7 +13,7 @@ exec := import("@milaboratory/tengo-sdk:exec")
 
 json := import("json")
 
-getPresetTpl := assets.importTemplate(":get-preset-for-back")
+getPresetTpl := assets.importTemplate(":get-preset")
 calculatePfconvParamsTpl := assets.importTemplate(":calculate-pfconv-params")
 processTpl := assets.importTemplate(":process")
 
@@ -33,10 +33,13 @@ wf.body(func(args) {
 			species: species
 		}
 	})
+
 	presetContent := getPreset.output("preset", 24 * 60 * 60 * 1000)
+	specsContent := getPreset.output("presetSpecForBack", 24 * 60 * 60 * 1000)
 
 	calculatePfconvParams := render.create(calculatePfconvParamsTpl, {
-		preset: presetContent
+		preset: presetContent,
+		specs: specsContent
 	})
 	pfconvParams := calculatePfconvParams.output("params", 24 * 60 * 60 * 1000)
 
@@ -45,7 +48,7 @@ wf.body(func(args) {
 			preset: preset,
 			species: species
 		}),
-		presetContent: presetContent,
+		presetContent: specsContent,
 		pfconvParams: pfconvParams,
 		inputSpec: input.getFutureInputField("spec"),
 		inputData: input.getFutureInputField("data")

diff --git a/workflow/src/mixcr-analyze.tpl.tengo b/workflow/src/mixcr-analyze.tpl.tengo
@@ -7,6 +7,7 @@ exec := import("@milaboratory/tengo-sdk:exec")
 smart := import("@milaboratory/tengo-sdk:smart")
 assets := import("@milaboratory/tengo-sdk:assets")
 render := import("@milaboratory/tengo-sdk:render")
+mixcr := import(":mixcr")
 
 json := import("json")
 
@@ -58,6 +59,8 @@ self.body(func(inputs) {
 							software(mixcrSw).
 							arg("analyze")
 
+		mixcr.addAdditionalColumnArgs(mixcrCmdBuilder)
+
 		if !is_undefined(species) {
 			mixcrCmdBuilder.arg("--species").arg(species)
 		}

diff --git a/workflow/src/mixcr.lib.tengo b/workflow/src/mixcr.lib.tengo
@@ -0,0 +1,18 @@
+ll := import("@milaboratory/tengo-sdk:ll")
+
+addAdditionalColumnArgs := func(mixcrCmdBuiler) {
+	mixcrCmdBuiler.
+		arg("--append-export-clones-field").
+		arg("-vHit").
+		arg("--append-export-clones-field").
+		arg("-dHit").
+		arg("--append-export-clones-field").
+		arg("-jHit").
+		arg("--append-export-clones-field").
+		arg("-cHit")
+	return mixcrCmdBuiler
+}
+
+export ll.toStrict({
+	addAdditionalColumnArgs: addAdditionalColumnArgs
+})