rename columns, add count by tag, option to don't show secondary chain

aglucaci · Sep 5, 2023 · 0e1d7bf · 0e1d7bf
1 parent 490479c
commit 0e1d7bf
Show file tree

Hide file tree

Showing 4 changed files with 40 additions and 22 deletions.
diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/CommandExportCloneGroups.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/CommandExportCloneGroups.kt
@@ -23,8 +23,6 @@ import com.milaboratory.mixcr.basictypes.CloneSetIO
 import com.milaboratory.mixcr.basictypes.IOUtil
 import com.milaboratory.mixcr.basictypes.MiXCRFileInfo
 import com.milaboratory.mixcr.basictypes.tag.TagCountAggregator
-import com.milaboratory.mixcr.basictypes.tag.TagType
-import com.milaboratory.mixcr.basictypes.tag.TagsInfo
 import com.milaboratory.mixcr.cli.CommonDescriptions.DEFAULT_VALUE_FROM_PRESET
 import com.milaboratory.mixcr.clonegrouping.CellType
 import com.milaboratory.mixcr.export.CloneGroup
@@ -162,10 +160,13 @@ object CommandExportCloneGroups {
             )
 
             val recalculatedClones = filterClones(initialSet, params, fileInfo)
-            val groups = cloneGroups(recalculatedClones, tagsInfo)
+            val groups = cloneGroups(recalculatedClones)
 
             val headerForExport = MetaForExport(fileInfo)
-            val fieldExtractors = CloneGroupFieldsExtractorsFactory(params.types.ifEmpty { CellType.values().toList() })
+            val fieldExtractors = CloneGroupFieldsExtractorsFactory(
+                params.types.ifEmpty { CellType.values().toList() },
+                params.showSecondaryChains
+            )
                 .createExtractors(params.fields, headerForExport)
 
             val rowMetaForExport = RowMetaForExport(
@@ -186,16 +187,10 @@ object CommandExportCloneGroups {
             }
         }
 
-        private fun cloneGroups(
-            recalculatedClones: CloneSet,
-            tagsInfo: TagsInfo
-        ): List<CloneGroup> = recalculatedClones
+        private fun cloneGroups(recalculatedClones: CloneSet): List<CloneGroup> = recalculatedClones
             .groupBy { it.group!! }
             .map { (groupId, clones) ->
                 val clonesGroupedByChains = clones.groupBy { it.chains }
-                val totalTagsCount = TagCountAggregator().also { aggregator ->
-                    clones.forEach { clone -> aggregator.add(clone.tagCount) }
-                }.createAndDestroy()
                 CloneGroup(
                     groupId = groupId,
                     clonePairs = clonesGroupedByChains.mapValues { (_, clonesWithChain) ->
@@ -205,10 +200,11 @@ object CommandExportCloneGroups {
                             sorted.getOrNull(1)
                         )
                     },
-                    cellsCount = totalTagsCount.getTagDiversity(tagsInfo.getDepthFor(TagType.Cell)),
                     cloneCount = clonesGroupedByChains.mapValues { it.value.size },
                     totalReadsCount = clones.sumOf { it.count },
-                    totalTagsCount = totalTagsCount
+                    totalTagsCount = TagCountAggregator().also { aggregator ->
+                        clones.forEach { clone -> aggregator.add(clone.tagCount) }
+                    }.createAndDestroy()
                 )
             }
             .sortedBy { it.groupId }

diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/Mixins.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/Mixins.kt
@@ -609,11 +609,29 @@ object ExportMiXCRMixins {
         @Option(
             description = ["Export clone groups for all cell types."],
             names = [ExportMixins.ExportCloneGroupsForAllCellTypes.CMD_OPTION],
-            order = OptionsOrder.mixins.exports + 5_200,
+            order = OptionsOrder.mixins.exports + 5_101,
             arity = "0",
         )
         fun exportCloneGroupsForAllCellTypes(ignored: Boolean) =
             mixIn(ExportMixins.ExportCloneGroupsForAllCellTypes)
+
+        @Option(
+            description = ["Show columns for secondary chains in export for cell groups."],
+            names = [ExportMixins.ShowSecondaryChainOnExportCellGroups.CMD_OPTION_FOR_SHOW],
+            order = OptionsOrder.mixins.exports + 5_200,
+            arity = "0",
+        )
+        fun showSecondaryChainOnExportCellGroups(ignored: Boolean) =
+            mixIn(ExportMixins.ShowSecondaryChainOnExportCellGroups(true))
+
+        @Option(
+            description = ["Don't show columns for secondary chains in export for cell groups."],
+            names = [ExportMixins.ShowSecondaryChainOnExportCellGroups.CMD_OPTION_FOR_DONT_SHOW],
+            order = OptionsOrder.mixins.exports + 5_201,
+            arity = "0",
+        )
+        fun dontShowSecondaryChainOnExportCellGroups(ignored: Boolean) =
+            mixIn(ExportMixins.ShowSecondaryChainOnExportCellGroups(false))
     }
 
     private interface ExportClonesMixins : GeneralExportClonesMixins {

diff --git a/src/main/resources/presets/blocks/10-exportCloneGroups.yaml b/src/main/resources/presets/blocks/10-exportCloneGroups.yaml
@@ -2,12 +2,13 @@ exportCloneGroups-single-cell-no-umi-base:
   abstract: true
   exportCloneGroups:
     types: [ ]
+    showSecondaryChains: true
     fields:
       - field: "-cellGroup"
-      - field: "-groupCellsCount"
-      - field: "-groupReadsCount"
+      - field: "-groupReadCount"
+      - field: "-groupUniqueTagCount"
+        args: [ "Cell" ]
       - field: "-uniqClonesPerChainCount"
-      - field: "-cloneId"
       - field: "-readCount"
       - field: "-targetSequences"
       - field: "-targetQualities"
@@ -27,12 +28,15 @@ exportCloneGroups-single-cell-with-umi-base:
   abstract: true
   exportCloneGroups:
     types: [ ]
+    showSecondaryChains: true
     fields:
       - field: "-cellGroup"
-      - field: "-groupCellsCount"
-      - field: "-groupReadsCount"
+      - field: "-groupReadCount"
+      - field: "-groupUniqueTagCount"
+        args: [ "Cell" ]
+      - field: "-groupUniqueTagCount"
+        args: [ "Molecule" ]
       - field: "-uniqClonesPerChainCount"
-      - field: "-cloneId"
       - field: "-readCount"
       - field: "-uniqueTagCount"
         args: [ "Molecule" ]

diff --git a/src/test/kotlin/com/milaboratory/mixcr/PresetsTest.kt b/src/test/kotlin/com/milaboratory/mixcr/PresetsTest.kt
@@ -193,14 +193,14 @@ class PresetsTest {
                     val bundle = Presets.MiXCRBundleResolver.resolvePreset(presetName)
                     MiXCRCommandDescriptor.groupClones in bundle.pipeline!!.steps
                 }
-                .forEach { presetName ->
+                .forAll { presetName ->
                     val bundle = Presets.MiXCRBundleResolver.resolvePreset(presetName)
                     val exportClonesFields = bundle.exportClones!!.fields
                     val exportCloneGroupsFields = bundle.exportCloneGroups!!.fields
 
                     presetName.asClue {
                         exportCloneGroupsFields shouldContainAll exportClonesFields
-                            .filter { it.field !in CloneGroupFieldsExtractorsFactory.excludeCloneFields }
+                            .filter { it.field !in (CloneGroupFieldsExtractorsFactory.excludeCloneFields + "-cloneId") }
                         exportCloneGroupsFields.forAll {
                             it.field shouldNotBeIn (CloneGroupFieldsExtractorsFactory.excludeCloneFields - "-cellGroup")
                         }