Merge pull request #1811 from milaboratory/bruker-wip

changed bruker preset to species agnostic, added cell id barcode tran…
milaboratory · Oct 7, 2024 · f703f29 · f703f29
2 parents 8b1ab7f + 71d0153
commit f703f29
Show file tree

Hide file tree

Showing 3 changed files with 80 additions and 8 deletions.
diff --git a/regression/presets/analyze/bruker-sc-xcr-vdj-beacon.yaml b/regression/presets/analyze/bruker-sc-xcr-vdj-beacon.yaml
@@ -1,4 +1,5 @@
-flags: []
+flags:
+  - species
 pipeline:
   - mitool-parse
   - mitool-refine-tags
@@ -118,7 +119,7 @@ mitool:
       safeDiscardedReads: true
       inferMinRecordsPerConsensus: true
 align:
-  species: hsa
+  species: ""
   libraryName: default
   trimmingQualityThreshold: 0
   trimmingWindowSize: 6
@@ -132,6 +133,40 @@ align:
   tagMaxBudget: 10.0
   headerExtractors: []
   readIdAsCellTag: false
+  tagTransformationSteps:
+    - type: mapTags
+      name: CELL to Bruker Barcode
+      matchTags:
+        - CELL
+      newTags:
+        - CELLBARCODE
+      allowOneMismatch: true
+      rules:
+        - from: ACCTACCGCC
+          to: C1D1F0T0
+        - from: ACTACTTCTG
+          to: C1D0F1T0
+        - from: AGACTCGATT
+          to: C0D1F0T1
+        - from: AGCCCTATCA
+          to: C0D1F0T0
+        - from: CCAGGATTAA
+          to: C0D1F1T1
+        - from: CTCCTTCAAG
+          to: C1D0F0T1
+        - from: GCCTTGTTGT
+          to: C1D1F1T1
+        - from: GTATACATTA
+          to: C1D0F0T0
+        - from: GTTAGCTGCT
+          to: C0D1F1T0
+        - from: TACATAAAGA
+          to: C0D0F0T1
+        - from: TCTCCAAGAC
+          to: C0D0F1T1
+        - from: TGGTAGGCTG
+          to: C0D0F1T0
+      reportingLevel: VerboseReport
   sampleTable: null
   tagsValidations:
     - type: MustContainTagType
@@ -277,6 +312,7 @@ align:
     saveOriginalReads: false
     smartForceEdgeAlignments: true
     retriesForPartialAlignments: 3
+    retriesAlignmentOfNotCoveredPart: 2
 refineTagsAndSort:
   whitelists: {}
   runCorrection: false

diff --git a/regression/presets/list.txt b/regression/presets/list.txt
@@ -220,6 +220,11 @@ takara-mouse-rna-bcr-umi-smarseq (SMART-Seq Mouse BCR (with UMIs))
 
 takara-human-rna-bcr-umi-smartseq (SMART-Seq Human BCR (with UMIs))
 
+-----Bruker-----
+bruker-sc-xcr-vdj-beacon (Single cell VDJ)
+  Required args:
+    --species <name>
+
 -----MiLaboratories-----
 milab-human-rna-tcr-umi-multiplex (Human TCR RNA Multiplex)
 
@@ -324,9 +329,6 @@ neb-mouse-rna-xcr-umi-nebnext (NEBNext® Immune Sequencing Kit (Mouse) BCR & TCR
   Required args:
     --species <name>
 
------Bruker-----
-bruker-human-sc-xcr-vdj-beacon (Single cell VDJ)
-
 -----AbHelix-----
 abhelix-human-rna-xcr (Human immunoglobulin repertoire)
 

diff --git a/src/main/resources/presets/protocols/bruker.yaml b/src/main/resources/presets/protocols/bruker.yaml
@@ -1,8 +1,10 @@
-"bruker-human-sc-xcr-vdj-beacon":
+"bruker-sc-xcr-vdj-beacon":
   vendor: "Bruker"
   label: "Single cell VDJ"
   category: non-generic
   inheritFrom: generic-lt-single-cell-amplicon
+  flags:
+    - species
   mixins:
     - type: SetTagPattern
       tagPattern: ^(CELL:N{10})ggg\^(R2:*)
@@ -12,8 +14,6 @@
       step: mitool-refine-tags
     - type: AddPipelineStep
       step: mitool-consensus
-    - type: SetSpecies
-      species: hsa
     - type: SetClonotypeAssemblingFeatures
       features: [ "VDJRegion" ]
     - type: LeftAlignmentBoundaryNoPoint
@@ -81,6 +81,40 @@
           parameters:
             mapperRelativeMinScore: 0.7
             relativeMinScore: 0.7
+      tagTransformationSteps:
+        - type: mapTags
+          name: "CELL to Bruker Barcode"
+          matchTags:
+            - CELL
+          newTags:
+            - CELLBARCODE
+          allowOneMismatch: true
+          rules:
+            - from: TGGTAGGCTG
+              to: C0D0F1T0
+            - from: GTTAGCTGCT
+              to: C0D1F1T0
+            - from: TACATAAAGA
+              to: C0D0F0T1
+            - from: AGCCCTATCA
+              to: C0D1F0T0
+            - from: ACCTACCGCC
+              to: C1D1F0T0
+            - from: TCTCCAAGAC
+              to: C0D0F1T1
+            - from: GTATACATTA
+              to: C1D0F0T0
+            - from: AGACTCGATT
+              to: C0D1F0T1
+            - from: CCAGGATTAA
+              to: C0D1F1T1
+            - from: CTCCTTCAAG
+              to: C1D0F0T1
+            - from: ACTACTTCTG
+              to: C1D0F1T0
+            - from: GCCTTGTTGT
+              to: C1D1F1T1
+          reportingLevel: VerboseReport
   refineTagsAndSort:
     runCorrection: false
     parameters: