Merge branch 'CW-4520' into 'dev'

don't draw SeqSummary plots if there were no reads after filtering [CW-4520]

Closes CW-4520

See merge request epi2melabs/workflows/wf-amplicon!91

.gitlab-ci.yml

@@ -33,6 +33,7 @@ docker-run:
             "ref-filter-all",
             "de-novo",
             "de-novo-spoa-max-len-2000",  # CW-3864
+            "de-novo-filter-all",  # CW-4520
             "ref-no-reads",
             "ref-single-read",
             "de-novo-no-reads",
@@ -115,7 +116,6 @@ docker-run:
         AFTER_NEXTFLOW_CMD: |
           ls -p $$CI_PROJECT_NAME | grep -q barcode01/ &&
           ls -p $$CI_PROJECT_NAME | grep -q barcode02/
-
     - if: $MATRIX_NAME == "de-novo-spoa-max-len-2000"
       variables:
         # we run with a higher `--force_spoa_length_threshold` so that the SPOA
@@ -136,6 +136,21 @@ docker-run:
         AFTER_NEXTFLOW_CMD: |
           ls -p $$CI_PROJECT_NAME | grep -q barcode01/ &&
           ls -p $$CI_PROJECT_NAME | grep -vq barcode02/
+    - if: $MATRIX_NAME == "de-novo-filter-all"
+      # this tests that no de-novo QC section was added to the report when there
+      # were no reads left after filtering
+      variables:
+        NF_WORKFLOW_OPTS: >
+          -executor.\$$local.memory 16GB
+          --fastq test_data/fastq-denovo
+          --max_read_length 1
+          --combine_results
+        NF_PROCESS_FILES: >
+          main.nf
+          lib/common.nf
+          modules/local/de-novo.nf
+        NF_IGNORE_PROCESSES: catFastqIntoFasta,concatTSVs,configure_igv,downsampleReads,medakaStitch,miniasm,mosdepthPerBase,mosdepthWindows,racon,spoa,trimAndQC
+        AFTER_NEXTFLOW_CMD: "! grep 'Quality Control' $$PWD/$$CI_PROJECT_NAME/*html"
 
     - if: $MATRIX_NAME == "ref-no-reads"
       variables:

CHANGELOG.md

@@ -8,6 +8,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Fixed
 - The workflow failing in _de novo_ mode when all reads for a sample came from the same strand.
 
+### Changed
+- The report now skips read length histograms etc. if there were no reads left after filtering.
+
 ## [v1.1.1]
 ### Changed
 - Updated Medaka to v1.12.0.

bin/workflow_glue/report.py

@@ -169,7 +169,7 @@ def populate_report(report, metadata, all_datasets, ref_fasta, downsampling_size
         # omit all other sections
         preprocessing_section(report, bad_datasets)
 
-        if de_novo:
+        if de_novo and bad_datasets_failed_qc:
             de_novo_qc_section(report, bad_datasets_failed_qc)
 
         return
@@ -617,6 +617,9 @@ def preprocessing_section(report, datasets):
     ]
     # there could be NaNs in case some datasets didn't have any reads
     preprocessing_stats.fillna(0, inplace=True)
+    # check if there were no reads at all after filtering + trimming (in which case
+    # we'll skip the `SeqSummary` plots)
+    no_reads_in_data = preprocessing_stats.loc["Filtered", "reads"] == 0
     # some formatting
     for col in ("reads", "bases"):
         preprocessing_stats[col] = (
@@ -642,9 +645,8 @@ def preprocessing_section(report, datasets):
         )
         DataTable.from_pandas(preprocessing_stats)
 
-        # return early if there were no per-read stats (i.e. there were no reads left
-        # after filtering)
-        if preprocessing_stats.loc["Filtered", "Reads"] == 0:
+        # return early if there were no reads left after filtering
+        if no_reads_in_data:
             return
         # `SeqSummary` plots
         html_tags.p(