Merge pull request #1169 from fls-bioinformatics-core/10x-ocm-data-add-qc-protocol

Update 10x Chromium 3' OCM application support and add new QC protocol

Author: pjbriggs

auto_process_ngs/applications.py

@@ -136,22 +136,23 @@
         "assays": ["10x Chromium 3' (v4 GEM-X) sc GEX",],
         "tags": ["10x", "single_cell"]
     },
-    # 10x Chromium 3' OCM single-cell
+    # 10x Chromium 3' OCM single-cell & single-nuclei
     {
         "platforms": ["10x Chromium 3' OCM"],
-        "libraries": ["scRNA-seq"],
+        "libraries": ["scRNA-seq", "snRNA-seq"],
         "extensions": ["CSP", "CRISPR"],
         "alternative_extensions": {
             "Antibody Capture": "CSP",
             "Feature Barcode": "CSP",
         },
         "fastq_generation": "10x_chromium_sc",
-        "qc_protocol": "10x_scRNAseq",
+        "qc_protocol": "10x_OCM",
         "setup": {
             "templates": ["10x_multi_config(multiplexing=ocm)"],
             "directories": [],
         },
-        "assays": ["10x Chromium 3' (v4 GEM-X) OCM sc GEX",],
+        "assays": ["10x Chromium 3' (v4 GEM-X) OCM sc GEX",
+                   "10x Chromium 3' (v4 GEM-X) OCM sn GEX",],
         "tags": ["10x", "single_cell", "multiplex"]
     },
     # 10x Chromium 3' single-nuclei
@@ -167,19 +168,6 @@
         "assays": ["10x Chromium 3' (v4 GEM-X) sn GEX",],
         "tags": ["10x", "single_cell"]
     },
-    # 10x Chromium 3' OCM single-nuclei
-    {
-        "platforms": ["10x Chromium 3' OCM"],
-        "libraries": ["snRNA-seq"],
-        "fastq_generation": "10x_chromium_sc",
-        "qc_protocol": "10x_snRNAseq",
-        "setup": {
-            "templates": ["10x_multi_config(multiplexing=ocm)"],
-            "directories": [],
-        },
-        "assays": ["10x Chromium 3' (v4 GEM-X) OCM sn GEX",],
-        "tags": ["10x", "single_cell", "multiplex"]
-    },
     # 10x Chromium 5'
     {
         "platforms": ["10x Chromium 5'"],

auto_process_ngs/mock.py

@@ -1877,7 +1877,7 @@ def create(path,exit_code=0,missing_fastqs=None,
             'ATAC' (when mocking 'cellranger-arc')
           multi_outputs (str): set type of
             outputs for 'cellranger multi' (either
-            'cellplex' or 'flex')
+            'ocm', 'flex' or 'cellplex')
           version (str): version of package to
             report
         """
@@ -2491,14 +2491,16 @@ def main(self,args):
                 with open(web_summary_file,'wt') as fp:
                     fp.write("PLACEHOLDER FOR WEB_SUMMARY.HTML")
             # Multiplexing outs
-            if self._multi_outputs == 'cellplex':
+            if self._multi_outputs == 'ocm':
+                multi_outputs = ("cells_per_tag.json",)
+            elif self._multi_outputs == 'flex':
+                multi_outputs = ("cells_per_tag.json",
+                                 "frp_gem_barcode_overlap.csv")
+            elif self._multi_outputs == 'cellplex':
                 multi_outputs = ("assignment_confidence_table.csv",
                                  "cells_per_tag.json",
                                  "tag_calls_per_cell.csv",
                                  "tag_calls_summary.csv")
-            elif self._multi_outputs == 'flex':
-                multi_outputs = ("cells_per_tag.json",
-                                 "frp_gem_barcode_overlap.csv")
             else:
                 multi_outputs = ()
             for f in multi_outputs:

auto_process_ngs/qc/modules/cellranger_multi.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 #
 #     cellranger_multi: implements 'cellranger_multi' QC module
-#     Copyright (C) University of Manchester 2024-2025 Peter Briggs
+#     Copyright (C) University of Manchester 2024-2026 Peter Briggs
 
 """
 Implements the 'cellranger_multi' QC module:
@@ -737,10 +737,11 @@ def finish(self):
                 qc_files = [f for f in expected_outputs(config_csv, multi_id)]
                 qc_files.append(os.path.join("outs",
                                              "config.csv"))
-                qc_files.append(os.path.join("outs",
-                                             "multi",
-                                             "multiplexing_analysis",
-                                             "tag_calls_summary.csv"))
+                for f in ("tag_calls_summary.csv",
+                          "cells_per_tag.json",):
+                    qc_files.append(os.path.join("outs",
+                                                 "multi",
+                                                 "multiplexing_analysis", f))
                 # Prefix dirs
                 prefix = multi_dir
                 if config.physical_sample:

auto_process_ngs/qc/protocols.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python
 #
 #     protocols: define and handle QC protocols
-#     Copyright (C) University of Manchester 2022-2025 Peter Briggs
+#     Copyright (C) University of Manchester 2022-2026 Peter Briggs
 #
 
 """
@@ -242,6 +242,26 @@ class in the ``check_outputs`` method; new modules must be added
         ]
     },
 
+    "10x_OCM": {
+        "description": "10xGenomics OCM (on-chip multiplexing) data",
+        "reads": {
+            "seq_data": ('r2',),
+            "index": ('r1',)
+        },
+        "qc_modules": [
+            'fastqc',
+            'fastq_screen',
+            'sequence_lengths',
+            'rseqc_genebody_coverage',
+            'rseqc_infer_experiment',
+            'qualimap_rnaseq',
+            'cellranger_count(cellranger_use_multi_config=True;'
+                             'set_cell_count=false;'
+                             'set_metadata=False)',
+            'cellranger_multi'
+        ]
+    },
+
     "10x_CellPlex": {
         "description": "10xGenomics CellPlex cell multiplexing data",
         "reads": {

auto_process_ngs/tenx/cellplex.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python
 #
 #     tenx/cellplex.py: utilities for handling 10xGenomics Cellplex data
-#     Copyright (C) University of Manchester 2023-2025 Peter Briggs
+#     Copyright (C) University of Manchester 2023-2026 Peter Briggs
 #
 
 """
@@ -145,7 +145,8 @@ def _read_config_csv(self, strict=True):
                     continue
                 if current_section == "samples":
                     if line.startswith('sample_id,cmo_ids') or \
-                       line.startswith('sample_id,probe_barcode_ids'):
+                       line.startswith('sample_id,probe_barcode_ids') or \
+                       line.startswith('sample_id,ocm_barcode_ids'):
                         # Header line, skip
                         continue
                     else:

auto_process_ngs/tenx/utils.py

@@ -543,7 +543,8 @@ def make_multi_config_template(f, reference=None, fastq_dir=None,
             MULTIPLEXED_SAMPLE,BC001|BC002|...,DESCRIPTION
             """))
         elif multiplexing == "ocm":
-            fp.write(dedent("""[samples]
+            fp.write(dedent("""
+            [samples]
             sample_id,ocm_barcode_ids,description
             MULTIPLEXED_SAMPLE,OB1|OB2|...,DESCRIPTION
             """))