Update cutnrun ci (#124)

* fix: add files to cutnrun ci

* lint: Snakefile

* fix: AV CI labels

* fix: controls with SEACR

* fix: check for blanks in chip2input

* bump: patch version

* Fix: run peakcaller-frip for all samples and added expand to frip_seacr rule.

---------

Co-authored-by: skchronicles <kuhnsa3@gmail.com>

Author: rroutsong

.github/workflows/analytical-validation-tests.yaml

@@ -184,6 +184,9 @@ jobs:
         /opt2/.tests/WT_S4_R1.001.fastq.gz /opt2/.tests/WT_S4_R2.001.fastq.gz \
         --output /opt2/output --peakcall /opt2/.tests/peakcall_no_replicates_no_inputs.tsv \
         --contrasts /opt2/.tests/contrasts.tsv --mode local --dry-run
+    - name: View the paired-end ATAC pipeline config file (without replicates, without inputs)
+      run: |
+        echo "Generated config file for ATAC pipeline (without replicates, without inputs)...." && jq -C '.' $PWD/output/config.json
     - name: Dry Run paired-end CUT & RUN pipeline, all options (without replicates, with inputs)
       run: |
         docker run -v $PWD:/opt2 snakemake/snakemake:v5.24.2 \
@@ -194,17 +197,22 @@ jobs:
         /opt2/.tests/Input_S4_R1.001.fastq.gz /opt2/.tests/Input_S4_R2.001.fastq.gz \
         --output /opt2/output --peakcall /opt2/.tests/peakcall_no_replicates_with_inputs.tsv \
         --contrasts /opt2/.tests/contrasts.tsv --mode local --dry-run
-    - name: Dry Run paired-end CUT & RUN pipeline, all options (without replicates, without inputs)
+    - name: View the paired-end CUT & RUN pipeline config file (without replicates, without inputs)
+      run: |
+        echo "Generated config file for CUT & RUN pipeline (without replicates, without inputs)...." && jq -C '.' $PWD/output/config.json
+    - name: Dry Run paired-end CUT & RUN pipeline all options, all options (without replicates, without inputs)
       run: |
         docker run -v $PWD:/opt2 snakemake/snakemake:v5.24.2 \
         /opt2/chrom-seek run --assay cutnrun --genome hg38  --input \
         /opt2/.tests/WT_S1.R1.fastq.gz /opt2/.tests/WT_S1.R2.fastq.gz \
         /opt2/.tests/WT_S4_R1.001.fastq.gz /opt2/.tests/WT_S4_R2.001.fastq.gz \
+        /opt2/.tests/Input_S1.R1.fastq.gz /opt2/.tests/Input_S1.R2.fastq.gz \
+        /opt2/.tests/Input_S4_R1.001.fastq.gz /opt2/.tests/Input_S4_R2.001.fastq.gz \
         --output /opt2/output --peakcall /opt2/.tests/peakcall_no_replicates_no_inputs.tsv \
         --contrasts /opt2/.tests/contrasts.tsv --mode local --dry-run
-    - name: View the paired-end ATAC pipeline config file (without replicates, without inputs)
+    - name: View the paired-end CUT & RUN pipeline config file (without replicates, without inputs)
       run: |
-        echo "Generated config file for ATAC pipeline (without replicates, without inputs)...." && jq -C '.' $PWD/output/config.json
+        echo "Generated config file for CUT & RUN pipeline (without replicates, without inputs)...." && jq -C '.' $PWD/output/config.json
     - name: Lint Workflow
       continue-on-error: true
       run: |

VERSION

@@ -1 +1 @@
-2.2.3
+2.2.4

workflow/Snakefile

@@ -151,9 +151,11 @@ rule_all_ins.extend(expand(
 rule_all_ins.append(join(workpath, "multiqc_report.html"))
 rule_all_ins.extend(expand(join(qc_dir, "{name}.preseq.dat"), name=samples))
 rule_all_ins.extend(
-    expand(join(peakqc_dir, "FRiP", "{PeakTool}", "{PeakTool}.{name}.Q5DD.FRiP_table.txt"), 
-    PeakTool=PeakTools, 
-    name=chips)
+    expand(
+        join(peakqc_dir, "FRiP", "{PeakTool}", "{PeakTool}.{name}.Q5DD.FRiP_table.txt"), 
+        PeakTool=PeakTools, 
+        name=samples
+    )
 )
 rule_all_ins.append(join(peakqc_dir, "FRiP", "summary.Q5DD.FRiP_heatmap.pdf"))
 rule_all_ins.extend(expand(

workflow/rules/paired/peakcall.smk

@@ -201,12 +201,11 @@ rule bam2bedgraph:
         bedtools genomecov -bg -i ${{tmp}}/sample.fragments.bed -g {params.reference} > {output}
         """
 
-
 rule SEACR:
     input:
         exp                             = join(bg_dir, "{name}.bedgraph"),
         control                         = lambda w: join(bg_dir, f"{chip2input[w.name]}.bedgraph")
-                                            if chip2input[w.name] and chip2input[w.name] != ""else [],
+                                            if w.name in chip2input and chip2input[w.name] != "" else [],
     output:
         peaks                           = join(seacr_dir, "{name}", "{name}.stringent.bed")
     params:

workflow/rules/peakqc.smk

@@ -126,7 +126,7 @@ rule FRiP_macsB:
 
 rule FRiP_SEACR:
     input:
-        peaks                   = join(seacr_dir, "{name}", "{name}.stringent.bed"),
+        peaks                   = expand(join(seacr_dir, "{name}", "{name}.stringent.bed"), name=chips),
         bam                     = join(bam_dir, "{name}.Q5DD.bam"),
     output:
         tbl                     = join(peakqc_dir, "FRiP", "SEACR", "SEACR.{name}.Q5DD.FRiP_table.txt"),