feat: --sample-table (SetSampleTable) mixin renamed to --sample-sheet…

… (SetSampleSheet) feat: case insensistive tag type deduction. Now "Cell" is valid tag name, not only "CELL". feat: tagsValidation in align fix: fixed grammar in PresetValidation
milaboratory · Jul 1, 2023 · 4b5e5d1 · 4b5e5d1
1 parent cc3c3b5
commit 4b5e5d1
Show file tree

Hide file tree

Showing 10 changed files with 65 additions and 40 deletions.
diff --git a/build.gradle.kts b/build.gradle.kts
@@ -130,7 +130,7 @@ val toObfuscate: Configuration by configurations.creating {
 
 val obfuscationLibs: Configuration by configurations.creating
 
-val mixcrAlgoVersion = "4.3.0-215-tiny-things"
+val mixcrAlgoVersion = "4.3.0-219-final-440-fixes"
 val milibVersion = "3.0.0-1-master"
 val mitoolVersion = ""
 val repseqioVersion = ""

diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/CommandAlign.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/CommandAlign.kt
@@ -998,6 +998,10 @@ object CommandAlign {
             // Tags
             val tagsExtractor = getTagsExtractor(cmdParams, inputFileGroups)
 
+            // Validating output tags if required
+            for (tagsValidation in cmdParams.tagsValidations)
+                tagsValidation.validate(tagsExtractor.tagsInfo)
+
             // true if final NSQTuple will have two reads, false otherwise
             val pairedPayload = tagsExtractor.pairedPatternPayload
                 ?: inputFileGroups.inputType.pairedRecords

diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/CommandAlignPipeline.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/CommandAlignPipeline.kt
@@ -183,24 +183,6 @@ object CommandAlignPipeline {
         )
     }
 
-    // fun inferSampleTable(fileGroups: CommandAlign.InputFileGroups): CommandAlignParams.SampleTable {
-    //     val sampleTagNames = fileGroups.tags.filter { TagType.detectByTagName(it) == TagType.Sample }
-    //     return CommandAlignParams.SampleTable(
-    //         sampleTagNames,
-    //         fileGroups.fileGroups
-    //             .map { fg -> sampleTagNames.map { fg.getTag(it) } }
-    //             .toSortedSet(listComparator())
-    //             .map { sample ->
-    //                 CommandAlignParams.SampleTable.Row(
-    //                     matchTags = sampleTagNames
-    //                         .mapIndexed { i, tn -> tn to sample[i] }
-    //                         .toMap(TreeMap()),
-    //                     sample = sample
-    //                 )
-    //             }
-    //     )
-    // }
-
     sealed interface TagExtractor {
         fun extract(
             originalReadId: Long,

diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/CommandListPresets.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/CommandListPresets.kt
@@ -57,6 +57,6 @@ private val flagOptions = mapOf(
             "${AlignMixins.AlignmentBoundaryConstants.RIGHT_RIGID_CMD_OPTION} [(${Labels.GENE_TYPE}|${Labels.ANCHOR_POINT})])",
     Flags.TagPattern to "${AlignMixins.SetTagPattern.CMD_OPTION} <pattern>",
     Flags.SampleTable to
-            "${AlignMixins.SetSampleTable.CMD_OPTION_FUZZY} sample_table.tsv\n" +
-            "${AlignMixins.SetSampleTable.CMD_OPTION_STRICT} sample_table.tsv",
+            "${AlignMixins.SetSampleSheet.CMD_OPTION_FUZZY} sample_table.tsv\n" +
+            "${AlignMixins.SetSampleSheet.CMD_OPTION_STRICT} sample_table.tsv",
 )
diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/MiXCRParamsResolver.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/MiXCRParamsResolver.kt
@@ -84,13 +84,13 @@ val presetFlagsMessages = mapOf(
     Flags.TagPattern to
             "This preset requires to specify tag pattern, \n" +
             "please use ${AlignMixins.SetTagPattern.CMD_OPTION} mix-in to set it, alternatively " +
-            "tag pattern can be provided with sample table using ${AlignMixins.SetSampleTable.CMD_OPTION_FUZZY} or " +
-            "${AlignMixins.SetSampleTable.CMD_OPTION_STRICT} mixin.",
+            "tag pattern can be provided with sample table using ${AlignMixins.SetSampleSheet.CMD_OPTION_FUZZY} or " +
+            "${AlignMixins.SetSampleSheet.CMD_OPTION_STRICT} mixin.",
 
     Flags.SampleTable to
             "This preset requires to specify sample table, \n" +
-            "please use ${AlignMixins.SetSampleTable.CMD_OPTION_FUZZY} or " +
-            "${AlignMixins.SetSampleTable.CMD_OPTION_STRICT} mix-in.",
+            "please use ${AlignMixins.SetSampleSheet.CMD_OPTION_FUZZY} or " +
+            "${AlignMixins.SetSampleSheet.CMD_OPTION_STRICT} mix-in.",
 )
 
 

diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/Mixins.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/Mixins.kt
@@ -183,23 +183,51 @@ class AlignMiXCRMixins : MiXCRMixinCollector() {
 
     @Option(
         description = ["Loads sample table from a tab separated file (one substitution will be allowed during matching)"],
-        names = [AlignMixins.SetSampleTable.CMD_OPTION_FUZZY],
+        names = [AlignMixins.SetSampleSheet.OLD_CMD_OPTION_FUZZY],
         arity = "1",
         paramLabel = "sample_table.tsv",
-        order = OptionsOrder.mixins.align + 330
+        hidden = true
     )
-    fun sampleTableFuzzy(arg: String) =
-        mixIn(AlignMixins.SetSampleTable(arg, null, true))
+    fun sampleTableFuzzy(arg: String) {
+        println(
+            "Option ${AlignMixins.SetSampleSheet.OLD_CMD_OPTION_FUZZY}is deprecated. Use ${AlignMixins.SetSampleSheet.CMD_OPTION_FUZZY} instead."
+        )
+        mixIn(AlignMixins.SetSampleSheet(arg, null, true))
+    }
+
+    @Option(
+        description = ["Loads sample table from a tab separated file (one substitution will be allowed during matching)"],
+        names = [AlignMixins.SetSampleSheet.CMD_OPTION_FUZZY],
+        arity = "1",
+        paramLabel = "sample_sheet.tsv",
+        order = OptionsOrder.mixins.align + 330,
+    )
+    fun sampleSheetFuzzy(arg: String) =
+        mixIn(AlignMixins.SetSampleSheet(arg, null, true))
 
     @Option(
         description = ["Loads sample table from a tab separated file (strict matching will be used)."],
-        names = [AlignMixins.SetSampleTable.CMD_OPTION_STRICT],
+        names = [AlignMixins.SetSampleSheet.OLD_CMD_OPTION_STRICT],
         arity = "1",
         paramLabel = "sample_table.tsv",
-        order = OptionsOrder.mixins.align + 331
+        hidden = true
+    )
+    fun sampleTableStrict(arg: String) {
+        println(
+            "Option ${AlignMixins.SetSampleSheet.OLD_CMD_OPTION_STRICT}is deprecated. Use ${AlignMixins.SetSampleSheet.CMD_OPTION_STRICT} instead."
+        )
+        mixIn(AlignMixins.SetSampleSheet(arg, null, false))
+    }
+
+    @Option(
+        description = ["Loads sample table from a tab separated file (strict matching will be used)."],
+        names = [AlignMixins.SetSampleSheet.CMD_OPTION_STRICT],
+        arity = "1",
+        paramLabel = "sample_sheet.tsv",
+        order = OptionsOrder.mixins.align + 335
     )
-    fun sampleTableStrict(arg: String) =
-        mixIn(AlignMixins.SetSampleTable(arg, null, false))
+    fun sampleSheetStrict(arg: String) =
+        mixIn(AlignMixins.SetSampleSheet(arg, null, false))
 
     //
     // Material type

diff --git a/src/main/resources/presets/blocks/01-align.yaml b/src/main/resources/presets/blocks/01-align.yaml
@@ -17,6 +17,13 @@ align-amplicon:
     tagUnstranded: false
     tagMaxBudget: 10
     splitBySample: false
+    tagsValidations:
+      # Note: the following validation is applicable to all built-in presets in all cases.
+      # Validation is performed on the set of output tags. When sample tags are employed and splitBySample is set
+      # to true, these tags are used within the alignment step and don't appear in the output files. Therefore,
+      # the validation remains relevant and valid.
+      - type: MustNotContainTagType
+        tagType: Sample
     parameters:
       fixSeed: true
       libraryStructure: Unknown

diff --git a/src/main/resources/presets/protocols/generic-amplicon.yaml b/src/main/resources/presets/protocols/generic-amplicon.yaml
@@ -45,6 +45,11 @@ generic-amplicon:
         upper: 0.05
 
 generic-amplicon-with-umi:
+  mixins:
+    - type: AddTagsValidation
+      validation:
+        type: MustContainTagType
+        tagType: Molecule
   inheritFrom: bundle-base
   flags:
     - species
@@ -57,9 +62,6 @@ generic-amplicon-with-umi:
     - refineTagsAndSort
     - assemble
     - exportClones
-  validation:
-    - type: TagPatternShouldContainsTagName
-      tagName: UMI
   align:
     inheritFrom: align-amplicon
   refineTagsAndSort:

diff --git a/src/main/resources/presets/test.yaml b/src/main/resources/presets/test.yaml
@@ -174,9 +174,11 @@
 "10x-vdj-tcr-qc-test":
   inheritFrom: 10x-vdj-tcr
   deprecation: "This preset is only for tests"
-  validation:
-    - type: TagPatternShouldContainsTagType
-      tagType: Cell
+  mixins:
+    - type: AddTagsValidation
+      validation:
+        type: MustContainTagType
+        tagType: Cell
   qc:
     checks:
       - type: OffTargetReads

diff --git a/src/test/kotlin/com/milaboratory/mixcr/MixinsTest.kt b/src/test/kotlin/com/milaboratory/mixcr/MixinsTest.kt
@@ -26,7 +26,7 @@ class MixinsTest {
                 "S2\t^attagaca \\ ^attacaca(CELL1:NNNN)\tACCC\n" +
                 "S3\t^attagaca \\ ^gacatata(CELL1:NNNN)\tATTG\n" +
                 "S4\t^attagaca \\ ^gacatata(CELL1:NNNN)\tACCC\n"
-        val slMixin = AlignMixins.SetSampleTable(null, sampleList)
+        val slMixin = AlignMixins.SetSampleSheet(null, sampleList)
         Assert.assertTrue(slMixin.packed)
         TestUtil.assertJson(K_YAML_OM, slMixin, false)
         val parsed = slMixin.parse()