fix postprocess of nodes and the same clones in the tree

milaboratory · Jun 20, 2022 · 81ccc85 · 81ccc85
1 parent 01b3e58
commit 81ccc85
Show file tree

Hide file tree

Showing 4 changed files with 38 additions and 29 deletions.
diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/BuildSHMTreeReport.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/BuildSHMTreeReport.kt
@@ -40,7 +40,7 @@ import kotlin.math.abs
 
 class BuildSHMTreeReport : AbstractCommandReport() {
     @get:JsonProperty("stepResults")
-    val stepResults: MutableList<StepResult> = mutableListOf()
+    val stepResults = mutableListOf<StepResult>()
     override fun getCommand() = CommandFindShmTrees.COMMAND_NAME
 
     fun onStepEnd(step: BuildSHMTreeStep, clonesWasAdded: Int, treesCountDelta: Int) {
@@ -197,6 +197,8 @@ class BuildSHMTreeReport : AbstractCommandReport() {
                 helper.writeField("Trees combined", -stepResult.treesCountDelta)
             }
         }
+        helper.writeField("Total trees count", stepResults.sumOf { it.treesCountDelta })
+        helper.writeField("Total clones count in trees", stepResults.sumOf { it.clonesWasAdded })
     }
 
     private fun printPages(

diff --git a/src/main/kotlin/com/milaboratory/mixcr/cli/CommandExportShmTreesNewick.kt b/src/main/kotlin/com/milaboratory/mixcr/cli/CommandExportShmTreesNewick.kt
@@ -18,7 +18,6 @@ import com.milaboratory.mixcr.trees.SHMTreesReader
 import com.milaboratory.mixcr.trees.forPostanalysis
 import io.repseq.core.VDJCLibraryRegistry
 import picocli.CommandLine.Command
-import picocli.CommandLine.Option
 import picocli.CommandLine.Parameters
 import kotlin.io.path.Path
 import kotlin.io.path.writeText
@@ -33,9 +32,6 @@ class CommandExportShmTreesNewick : ACommandWithOutputMiXCR() {
     @Parameters(arity = "2", description = ["input_file.hsmt output_dir"])
     var inOut: List<String> = ArrayList()
 
-    @Option(description = ["Output column headers with spaces."], names = ["-v", "--with-spaces"])
-    var humanReadable = false
-
     override fun getInputFiles(): List<String> = listOf(inOut.first())
 
     override fun getOutputFiles(): List<String> = listOf(inOut.last())

diff --git a/src/main/kotlin/com/milaboratory/mixcr/trees/ClusterProcessor.kt b/src/main/kotlin/com/milaboratory/mixcr/trees/ClusterProcessor.kt
@@ -349,7 +349,7 @@ internal class ClusterProcessor private constructor(
                 }
             }
         }
-        val notOverlappedCliques: MutableList<BitArrayInt> = ArrayList()
+        val notOverlappedCliques = mutableListOf<BitArrayInt>()
         matrix.calculateMaximalCliques()
             .filter { it.bitCount() > 1 }
             .sortedByDescending { it.bitCount() }
@@ -358,11 +358,16 @@ internal class ClusterProcessor private constructor(
                     notOverlappedCliques.add(clique)
                 }
             }
-        val clusters: MutableList<Cluster<CloneWithMutationsFromVJGermline>> = ArrayList()
-        for (clique in notOverlappedCliques) {
-            clusters.add(Cluster(clique.bits.map { index: Int -> clones[index] }))
-        }
-        return clusters
+        return notOverlappedCliques
+            .map { Cluster(it.bits.map { i -> clones[i] }) }
+            .filter {
+                //skip cluster if it formed by the same clones but with different C or from different samples
+                val toCompare = it.cluster.first()
+                it.cluster.subList(1, it.cluster.size)
+                    .any { clone ->
+                        !Arrays.equals(clone.cloneWrapper.clone.targets, toCompare.cloneWrapper.clone.targets)
+                    }
+            }
     }
 
     private fun commonMutationsCount(

diff --git a/src/main/kotlin/com/milaboratory/mixcr/trees/TreeBuilderByAncestors.kt b/src/main/kotlin/com/milaboratory/mixcr/trees/TreeBuilderByAncestors.kt
@@ -319,9 +319,8 @@ class TreeBuilderByAncestors<T, E, M> private constructor(
         private val sumOfDistancesFromAncestor: BigDecimal,
         private val distanceFromParentToReplaceWhat: BigDecimal
     ) : Action() {
-        override fun changeOfDistance(): BigDecimal {
-            return distanceFromParentToCommon.subtract(distanceFromParentToReplaceWhat).add(sumOfDistancesFromAncestor)
-        }
+        override fun changeOfDistance(): BigDecimal =
+            distanceFromParentToCommon - distanceFromParentToReplaceWhat + sumOfDistancesFromAncestor
 
         override fun distanceFromObserved(): BigDecimal {
             return (parent.content as Reconstructed<T, E>).minDistanceFromObserved
@@ -335,38 +334,45 @@ class TreeBuilderByAncestors<T, E, M> private constructor(
             )
         }
 
-        private fun postprocess(parentToPostprocess: Tree.Node<ObservedOrReconstructed<T, E>>): Tree.Node<ObservedOrReconstructed<T, E>> {
-            return Tree.Node(
+        private fun postprocess(parentToPostprocess: Tree.Node<ObservedOrReconstructed<T, E>>): Tree.Node<ObservedOrReconstructed<T, E>> =
+            Tree.Node(
                 parentToPostprocess.content,
                 postprocess(
                     (parentToPostprocess.content as Reconstructed<T, E>).content,
                     parentToPostprocess.links
                 )
             )
-        }
 
         private fun postprocess(
             parentContent: E, links: List<NodeLink<ObservedOrReconstructed<T, E>>>
         ): List<NodeLink<ObservedOrReconstructed<T, E>>> = links
-            .map { link ->
+            .flatMap { link ->
                 val child = link.node
                 when (child.content) {
                     is Reconstructed<*, *> -> {
                         val childContent = child.content as Reconstructed<T, E>
                         val mapped = postprocessDescendants(parentContent, childContent.content)
-                        NodeLink(
-                            Tree.Node(
-                                Reconstructed(
-                                    mapped,  //TODO recalculate minDistanceFromObserved
-                                    childContent.minDistanceFromObserved,
-                                    ++counter
+                        val newDistance = distance(parentContent, mutationsBetween(parentContent, mapped))
+                        if (newDistance.compareTo(BigDecimal.ZERO) == 0) {
+                            //actually parent equals mapped result.
+                            //in this case we should remove this node from tree by skipping it and inserting its children directly
+                            postprocess(parentContent, child.links)
+                        } else {
+                            val result = NodeLink(
+                                Tree.Node(
+                                    Reconstructed(
+                                        mapped,  //TODO recalculate minDistanceFromObserved
+                                        childContent.minDistanceFromObserved,
+                                        ++counter
+                                    ),
+                                    postprocess(mapped, child.links)
                                 ),
-                                postprocess(mapped, child.links)
-                            ),
-                            distance(parentContent, mutationsBetween(parentContent, mapped))
-                        )
+                                newDistance
+                            )
+                            listOf(result)
+                        }
                     }
-                    else -> link
+                    else -> listOf(link)
                 }
             }
     }