Fetch read length info for run instead of just lane 1 (#66)

* Fetch read length info for run instead of just lane 1

* updated actions/cache: v2 --> v.4.2.2

* reformatted with black

* add unit test for read_run_info()

* sort reads list by @Number

* removed printVersion function

* removed version information from manifest

* updated ubuntu version for running tests

* changed ubuntu to 22.04 instead

Author: kjellinjonas

.github/workflows/run_tests.yml

@@ -2,7 +2,7 @@ name: Run tests
 on: [push, pull_request]
 jobs:
   run-tests:
-    runs-on: ubuntu-20.04
+    runs-on: ubuntu-22.04
     env:
       NXF_VER: 21.04.1
       NXF_ANSI_LOG: false
@@ -11,7 +11,7 @@ jobs:
         uses: actions/checkout@v2
 
       - name: Cache singularity images
-        uses: actions/cache@v2
+        uses: actions/cache@v4.2.2
         with:
           path: work/singularity
           key: singularity-${{ hashFiles('config/nextflow_config/singularity.config') }}

bin/get_metadata.py

@@ -6,13 +6,15 @@
 import argparse
 import os
 import json
+from operator import itemgetter
 from pathlib import Path
 import yaml
 
 
 class RunfolderInfo:
     def __init__(self, runfolder, bcl2fastq_outdir):
         self.runfolder = runfolder
+        self.run_info = self.read_run_info()
         self.run_parameters = self.read_run_parameters()
         self.stats_json = self.read_stats_json(bcl2fastq_outdir)
         self.description_and_identifier = OrderedDict()
@@ -42,6 +44,14 @@ def find(self, d, tag):
                     for i in v:
                         yield from self.find(i, tag)
 
+    def read_run_info(self):
+        run_info = os.path.join(self.runfolder, "RunInfo.xml")
+        if os.path.exists(run_info):
+            with open(run_info) as f:
+                return xmltodict.parse(f.read())
+        else:
+            return None
+
     def read_run_parameters(self):
         alt_1 = os.path.join(self.runfolder, "runParameters.xml")
         alt_2 = os.path.join(self.runfolder, "RunParameters.xml")
@@ -109,18 +119,25 @@ def get_read_cycles(self):
         read_counter = 1
         index_counter = 1
         try:
-            for read_info in self.stats_json["ReadInfosForLanes"][0]["ReadInfos"]:
-                if read_info["IsIndexedRead"]:
+            reads = self.run_info["RunInfo"]["Run"]["Reads"]["Read"]
+
+            # Handle potential cases with only one Read element
+            if isinstance(reads, dict):
+                reads = [reads]  # Wrap it in a list to make it iterable
+
+            for read_info in sorted(reads, key=itemgetter("@Number")):
+                if read_info["@IsIndexedRead"] == "Y":
                     read_and_cycles[f"Index {index_counter} (bp)"] = read_info[
-                        "NumCycles"
+                        "@NumCycles"
                     ]
                     index_counter += 1
                 else:
                     read_and_cycles[f"Read {read_counter} (bp)"] = read_info[
-                        "NumCycles"
+                        "@NumCycles"
                     ]
                     read_counter += 1
             return read_and_cycles
+
         except TypeError:
             return read_and_cycles
 

main.nf

@@ -61,14 +61,6 @@ def helpMessage() {
     """
 }
 
-def printVersion() {
-
-    log.info "seqreports v${workflow.manifest.version}"
-
-}
-
-printVersion()
-
 if (params.help || !params.run_folder){
     helpMessage()
     exit 0

nextflow.config

@@ -5,7 +5,6 @@ manifest {
     description = 'A Nextflow run folder QC pipeline for SciLifeLab SNP&SEQ platform'
     mainScript = 'main.nf'
     nextflowVersion = '!>=21.04.1'
-    version = '1.2.0'
 }
 
 profiles {

tests/unit_tests/test_get_metadata.py

@@ -108,3 +108,8 @@ def test_get_read_cycles(runfolder_info):
 def test_get_info(runfolder_info):
     results = runfolder_info.get_info()
     assert len(results) == 9
+
+
+def test_read_run_info(runfolder_info):
+    run_info = runfolder_info.read_run_info()
+    assert len(run_info["RunInfo"]["Run"]["Reads"]["Read"]) == 4