@platforma-open/milaboratories.mixcr-shm-trees.workflow 3.2.1 → 3.3.1

package/.turbo/turbo-build.log

@@ -1,5 +1,5 @@
 
-> @platforma-open/milaboratories.mixcr-shm-trees.workflow@3.2.1 build /home/runner/work/mixcr-shm-trees/mixcr-shm-trees/workflow
+> @platforma-open/milaboratories.mixcr-shm-trees.workflow@3.3.1 build /home/runner/work/mixcr-shm-trees/mixcr-shm-trees/workflow
 > rm -rf dist && pl-tengo check && pl-tengo build
 
 Processing "src/export-settings.lib.tengo"...

package/CHANGELOG.md

@@ -1,5 +1,17 @@
 # @platforma-open/milaboratories.mixcr-shm-trees.workflow
 
+## 3.3.1
+
+### Patch Changes
+
+- d8d8f49: Additional ensure uniqueness step in trees reconstruction workflow
+
+## 3.3.0
+
+### Minor Changes
+
+- b8c818f: Alignment-based sequence search for sequence collections, multiple other improvements for this subsystem
+
 ## 3.2.1
 
 ### Patch Changes

package/dist/tengo/lib/soi-export.lib.tengo

@@ -1,6 +1,9 @@
 ll := import("@platforma-sdk/workflow-tengo:ll")
 
 soiResultImportColumns := func(dbParameters) {
+	commonDomain := {
+		"pl7.app/list": dbParameters.id
+	}
 	if dbParameters.searchParameters.type == "tree_search_top" {
 		return [ {
 			column: "topHit",
@@ -12,9 +15,7 @@ soiResultImportColumns := func(dbParameters) {
 					"pl7.app/label": "Top Hit " + dbParameters.name
 				}
 			},
-			domain: {
-				"pl7.app/list": dbParameters.id
-			}
+			domain: commonDomain
 		}, {
 			column: "mutations",
 			id: "mutations",
@@ -25,9 +26,42 @@ soiResultImportColumns := func(dbParameters) {
 					"pl7.app/label": "Number of mutations " + dbParameters.name
 				}
 			},
-			domain: {
-				"pl7.app/list": dbParameters.id
-			}
+			domain: commonDomain
+		} ]
+	} else if dbParameters.searchParameters.type == "preset_alignment_search_top" {
+		return [ {
+			column: "topHit",
+			id: "topHit",
+			spec: {
+				name: "pl7.app/search/topHit",
+				valueType: "String",
+				annotations: {
+					"pl7.app/label": "Top Hit " + dbParameters.name
+				}
+			},
+			domain: commonDomain
+		}, {
+			column: "penalty",
+			id: "penalty",
+			spec: {
+				name: "pl7.app/search/alignmentPenalty",
+				valueType: "Int",
+				annotations: {
+					"pl7.app/label": "Alignment penalty " + dbParameters.name
+				}
+			},
+			domain: commonDomain
+		}, {
+			column: "score",
+			id: "score",
+			spec: {
+				name: "pl7.app/search/alignmentScore",
+				valueType: "Int",
+				annotations: {
+					"pl7.app/label": "Alignment score " + dbParameters.name
+				}
+			},
+			domain: commonDomain
 		} ]
 	} else {
 		ll.panic("Unknown search mode: " + dbParameters.searchParameters.type)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@platforma-open/milaboratories.mixcr-shm-trees.workflow",
-  "version": "3.2.1",
+  "version": "3.3.1",
   "type": "module",
   "description": "Tengo-based template",
   "//": {
@@ -8,13 +8,13 @@
   },
   "devDependencies": {
     "@platforma-sdk/tengo-builder": "^1.17.5",
-    "@platforma-sdk/workflow-tengo": "^2.9.2",
-    "@milaboratories/software-pframes-conv": "^2.1.3",
+    "@platforma-sdk/workflow-tengo": "^2.9.6",
+    "@milaboratories/software-pframes-conv": "^2.1.5",
     "@platforma-open/milaboratories.software-small-binaries": "^1.15.6",
     "@platforma-open/milaboratories.software-mixcr": "4.7.0-139-develop",
-    "@platforma-open/milaboratories.software-mitool": "2.3.1-2-main",
+    "@platforma-open/milaboratories.software-mitool": "2.3.1-5-main",
     "@platforma-open/milaboratories.software-paggregate": "^1.0.1",
-    "@platforma-sdk/test": "^1.21.9",
+    "@platforma-sdk/test": "^1.21.20",
     "vitest": "^2.1.8",
     "typescript": "~5.6.3"
   },

package/src/process.tpl.tengo

@@ -122,6 +122,7 @@ self.body(func(inputs) {
 		{
 			shmTreeTableArgs: shmTreeTableOptions.cmdArgs,
 			shmTreeNodesTableOptions: shmTreeNodesTableOptions,
+			shmTreeNodesWithClonesTableOptions: shmTreeNodesWithClonesTableOptions,
 			shmTreeNodesWithClonesTableArgs: shmTreeNodesWithClonesTableOptions.cmdArgs,
 			globalParams: maps.merge(
 				inputs.params,

package/src/reconstruct-shm-trees.tpl.tengo

@@ -194,37 +194,40 @@ self.body(func(inputs) {
 
     // aggregating by-nodes output to make it uniquely addressable by it's native key
 
-    aggregations := []
-	for col in inputs.shmTreeNodesTableOptions.pfconvParams.columns {
-		aggregations = append(aggregations, {
-			type: "first",
-			src: col.column,
-			dst: col.column
-		})
-	}
-
-    keyColumns := []
-    for axis in inputs.shmTreeNodesTableOptions.pfconvParams.axes {
-		keyColumns = append(keyColumns, axis.column)
-	}
-
-	aggregationWorkflow := { steps: [ {
-			type: "aggregate",
-			groupBy: keyColumns,
-			aggregations: aggregations
-		} ] }
-
-	aggregateCmd := exec.builder().
-        printErrStreamToStdout().
-        software(paggregateSw).
-        arg("--workflow").arg("wf.json").
-		writeFile("wf.json", json.encode(aggregationWorkflow)).
-		arg("input.tsv").addFile("input.tsv", shmTreeNodesTsvRaw).
-		arg("output.tsv").saveFile("output.tsv").
-		run()
+    ensureUniqueness := func(inputTsv, pfConvParams) {
+        aggregations := []
+        for col in pfConvParams.columns {
+            aggregations = append(aggregations, {
+                type: "first",
+                src: col.column,
+                dst: col.column
+            })
+        }
+
+        keyColumns := []
+        for axis in pfConvParams.axes {
+            keyColumns = append(keyColumns, axis.column)
+        }
 
-	shmTreeNodesTsv := aggregateCmd.getFile("output.tsv")
+        aggregationWorkflow := { steps: [ {
+                type: "aggregate",
+                groupBy: keyColumns,
+                aggregations: aggregations
+            } ] }
+
+        aggregateCmd := exec.builder().
+            printErrStreamToStdout().
+            software(paggregateSw).
+            arg("--workflow").arg("wf.json").
+            writeFile("wf.json", json.encode(aggregationWorkflow)).
+            arg("input.tsv").addFile("input.tsv", inputTsv).
+            arg("output.tsv").saveFile("output.tsv").
+            run()
+
+        return aggregateCmd.getFile("output.tsv")
+    }
 
+	shmTreeNodesTsv := ensureUniqueness(shmTreeNodesTsvRaw, inputs.shmTreeNodesTableOptions.pfconvParams)
 
     // export nodes with clones. For each node could be several clones
     shmTreeNodesWithClonesExportsCmdBuilder := exec.builder().
@@ -248,7 +251,9 @@ self.body(func(inputs) {
         saveFile("output.tsv")
 
     shmTreeNodesWithClonesExports := shmTreeNodesWithClonesExportsCmdBuilder.run()
-    shmTreeNodesWithClonesTsv := shmTreeNodesWithClonesExports.getFile("output.tsv")
+    shmTreeNodesWithClonesTsvRaw := shmTreeNodesWithClonesExports.getFile("output.tsv")
+
+    shmTreeNodesWithClonesTsv := ensureUniqueness(shmTreeNodesWithClonesTsvRaw, inputs.shmTreeNodesWithClonesTableOptions.pfconvParams)
 
     return {
         trees: shmTreeTsv,

package/src/soi-export.lib.tengo

@@ -1,6 +1,9 @@
 ll := import("@platforma-sdk/workflow-tengo:ll")
 
 soiResultImportColumns := func(dbParameters) {
+	commonDomain := {
+		"pl7.app/list": dbParameters.id
+	}
 	if dbParameters.searchParameters.type == "tree_search_top" {
 		return [ {
 			column: "topHit",
@@ -12,9 +15,7 @@ soiResultImportColumns := func(dbParameters) {
 					"pl7.app/label": "Top Hit " + dbParameters.name
 				}
 			},
-			domain: {
-				"pl7.app/list": dbParameters.id
-			}
+			domain: commonDomain
 		}, {
 			column: "mutations",
 			id: "mutations",
@@ -25,9 +26,42 @@ soiResultImportColumns := func(dbParameters) {
 					"pl7.app/label": "Number of mutations " + dbParameters.name
 				}
 			},
-			domain: {
-				"pl7.app/list": dbParameters.id
-			}
+			domain: commonDomain
+		} ]
+	} else if dbParameters.searchParameters.type == "preset_alignment_search_top" {
+		return [ {
+			column: "topHit",
+			id: "topHit",
+			spec: {
+				name: "pl7.app/search/topHit",
+				valueType: "String",
+				annotations: {
+					"pl7.app/label": "Top Hit " + dbParameters.name
+				}
+			},
+			domain: commonDomain
+		}, {
+			column: "penalty",
+			id: "penalty",
+			spec: {
+				name: "pl7.app/search/alignmentPenalty",
+				valueType: "Int",
+				annotations: {
+					"pl7.app/label": "Alignment penalty " + dbParameters.name
+				}
+			},
+			domain: commonDomain
+		}, {
+			column: "score",
+			id: "score",
+			spec: {
+				name: "pl7.app/search/alignmentScore",
+				valueType: "Int",
+				annotations: {
+					"pl7.app/label": "Alignment score " + dbParameters.name
+				}
+			},
+			domain: commonDomain
 		} ]
 	} else {
 		ll.panic("Unknown search mode: " + dbParameters.searchParameters.type)

package/src/soi.tpl.tengo

@@ -1,4 +1,5 @@
 self := import("@platforma-sdk/workflow-tengo:tpl")
+ll := import("@platforma-sdk/workflow-tengo:ll")
 assets := import("@platforma-sdk/workflow-tengo:assets")
 maps := import("@platforma-sdk/workflow-tengo:maps")
 exec := import("@platforma-sdk/workflow-tengo:exec")
@@ -74,6 +75,58 @@ self.body(func(inputs) {
 
 	inputTsv := xsv.exportFrame([{spec: modifiedQuerySpec, data: queryData}], "tsv", {})
 
+	actualSearchParameters := db.parameters.searchParameters
+	if actualSearchParameters.type == "preset_alignment_search_top" {
+		// nVDJRegion: 369
+		// nCDR3: 51
+		// aaVDJRegion: 641
+		// aaCDR3: 93
+		if db.parameters.type == "nucleotide" {
+			actualSearchParameters = {
+				type: "banded_alignment_search_top",
+				scoring: {
+					match: 1,
+					mismatch: -2,
+					gap: {
+						type: "affine",
+						openPenalty: -3,
+						extensionPenalty: -1
+					}
+				},
+				band: 9
+			}
+			if db.parameters.targetFeature == "CDR3" {
+				actualSearchParameters.maxPenalty = 1 + int(51 * db.parameters.searchParameters.dissimilarityPercent / 100)
+			} else if db.parameters.targetFeature == "VDJRegion" {
+				actualSearchParameters.maxPenalty = 1 + int(369 * db.parameters.searchParameters.dissimilarityPercent / 100)
+			} else {
+				ll.panic("Unknown target feature: " + db.parameters.targetFeature)
+			}
+		} else if db.parameters.type == "amino-acid" {
+			actualSearchParameters = {
+				type: "banded_alignment_search_top",
+				scoring: {
+					matrix: "BLOSUM62",
+					gap: {
+						type: "affine",
+						openPenalty: -10,
+						extensionPenalty: -1
+					}
+				},
+				band: 4
+			}
+			if db.parameters.targetFeature == "CDR3" {
+				actualSearchParameters.maxPenalty = 1 + int(93 * db.parameters.searchParameters.dissimilarityPercent / 100)
+			} else if db.parameters.targetFeature == "VDJRegion" {
+				actualSearchParameters.maxPenalty = 1 + int(641 * db.parameters.searchParameters.dissimilarityPercent / 100)
+			} else {
+				ll.panic("Unknown target feature: " + db.parameters.targetFeature)
+			}
+		} else {
+			ll.panic("Unknown alphabet: " + db.parameters.type)
+		}
+	}
+
 	searchCmd := exec.builder().
         printErrStreamToStdout().
         secret("MI_LICENSE", "MI_LICENSE").
@@ -83,7 +136,7 @@ self.body(func(inputs) {
 		arg("--database").arg("database.tsv").
 		writeFile("database.tsv", dbData).
 		arg("--parameters").arg("params.json").
-		writeFile("params.json", json.encode(db.parameters.searchParameters)).
+		writeFile("params.json", json.encode(actualSearchParameters)).
 		arg("--hits-only").
 		arg("--target-column").arg("query").
         arg("input.tsv").addFile("input.tsv", inputTsv).