Pool inputs

CHANGELOG.md

@@ -1,6 +1,7 @@
 ## CHAMPAGNE development version
 
 - The CHAMPAGNE nextflow workflow now has a version entry in `nextflow.config`, in compliance with nf-core. (#213, @kelly-sovacool)
+- Pool input (control) reads of the same sample name by default. Any inputs that should not be pooled must have different sample names in the samplesheet. (#214, @kelly-sovacool)
 
 ## CHAMPAGNE 0.4.0
 

assets/samplesheet_human.csv

@@ -1,13 +1,13 @@
 fastq_1,fastq_2,sample,rep,antibody,control
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129678.fastq.gz,,A549_CTCF,1,CTCF,A549_CTCF_INPUT_1
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129676.fastq.gz,,A549_CTCF,2,CTCF,A549_CTCF_INPUT_2
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129677.fastq.gz,,A549_CTCF,3,CTCF,A549_CTCF_INPUT_3
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129560.fastq.gz,,A549_CTCF_INPUT_1,,,
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129561.fastq.gz,,A549_CTCF_INPUT_2,,,
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129562.fastq.gz,,A549_CTCF_INPUT_3,,,
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14636612.fastq.gz,,A549_JUN,1,JUN,A549_JUN_INPUT_1
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14636613.fastq.gz,,A549_JUN,2,JUN,A549_JUN_INPUT_2
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14636614.fastq.gz,,A549_JUN,3,JUN,A549_JUN_INPUT_2
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14638304.fastq.gz,,A549_JUN_INPUT_1,,,
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14638305.fastq.gz,,A549_JUN_INPUT_2,,,
-/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14638306.fastq.gz,,A549_JUN_INPUT_3,,,
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129678.fastq.gz,,A549_CTCF,1,CTCF,A549_CTCF_INPUT
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129676.fastq.gz,,A549_CTCF,2,CTCF,A549_CTCF_INPUT
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129677.fastq.gz,,A549_CTCF,3,CTCF,A549_CTCF_INPUT
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129560.fastq.gz,,A549_CTCF_INPUT,1,,
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129561.fastq.gz,,A549_CTCF_INPUT,2,,
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR5129562.fastq.gz,,A549_CTCF_INPUT,3,,
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14636612.fastq.gz,,A549_JUN,1,JUN,A549_JUN_INPUT
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14636613.fastq.gz,,A549_JUN,2,JUN,A549_JUN_INPUT
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14636614.fastq.gz,,A549_JUN,3,JUN,A549_JUN_INPUT
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14638304.fastq.gz,,A549_JUN_INPUT,1,,
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14638305.fastq.gz,,A549_JUN_INPUT,2,,
+/data/CCBR_Pipeliner/testdata/CHAMPAGNE/human/SRR14638306.fastq.gz,,A549_JUN_INPUT,3,,

assets/samplesheet_test.csv

@@ -1,7 +1,7 @@
 sample,rep,fastq_1,fastq_2,antibody,control
-SPT5_T0,1,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822153_1.fastq.gz,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822153_2.fastq.gz,SPT5,SPT5_INPUT_1
-SPT5_T0,2,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822154_1.fastq.gz,,SPT5,SPT5_INPUT_2
-SPT5_T15,1,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822157_1.fastq.gz,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822157_2.fastq.gz,SPT5,SPT5_INPUT_1
-SPT5_T15,2,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822158_1.fastq.gz,,SPT5,SPT5_INPUT_2
-SPT5_INPUT_1,,https://raw.githubusercontent.com/nf-core/test-datasets/chipseq/testdata/SRR5204809_Spt5-ChIP_Input1_SacCer_ChIP-Seq_ss100k_R1.fastq.gz,https://raw.githubusercontent.com/nf-core/test-datasets/chipseq/testdata/SRR5204809_Spt5-ChIP_Input1_SacCer_ChIP-Seq_ss100k_R2.fastq.gz,,
-SPT5_INPUT_2,,https://raw.githubusercontent.com/nf-core/test-datasets/chipseq/testdata/SRR5204810_Spt5-ChIP_Input2_SacCer_ChIP-Seq_ss100k_R1.fastq.gz,,,
+SPT5_T0,1,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822153_1.fastq.gz,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822153_2.fastq.gz,SPT5,SPT5_INPUT
+SPT5_T0,2,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822154_1.fastq.gz,,SPT5,SPT5_INPUT
+SPT5_T15,1,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822157_1.fastq.gz,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822157_2.fastq.gz,SPT5,SPT5_INPUT
+SPT5_T15,2,https://raw.githubusercontent.com/nf-core/test-datasets/atacseq/testdata/SRR1822158_1.fastq.gz,,SPT5,SPT5_INPUT
+SPT5_INPUT,1,https://raw.githubusercontent.com/nf-core/test-datasets/chipseq/testdata/SRR5204809_Spt5-ChIP_Input1_SacCer_ChIP-Seq_ss100k_R1.fastq.gz,https://raw.githubusercontent.com/nf-core/test-datasets/chipseq/testdata/SRR5204809_Spt5-ChIP_Input1_SacCer_ChIP-Seq_ss100k_R2.fastq.gz,,
+SPT5_INPUT,2,https://raw.githubusercontent.com/nf-core/test-datasets/chipseq/testdata/SRR5204810_Spt5-ChIP_Input2_SacCer_ChIP-Seq_ss100k_R1.fastq.gz,https://raw.githubusercontent.com/nf-core/test-datasets/chipseq/testdata/SRR5204810_Spt5-ChIP_Input2_SacCer_ChIP-Seq_ss100k_R2.fastq.gz,,

bin/check_samplesheet.py

@@ -1,9 +1,9 @@
 #!/usr/bin/env python3
 
 """
-source: https://github.com/nf-core/chipseq/blob/51eba00b32885c4d0bec60db3cb0a45eb61e34c5/bin/check_samplesheet.py
+adapted from: https://github.com/nf-core/chipseq/blob/51eba00b32885c4d0bec60db3cb0a45eb61e34c5/bin/check_samplesheet.py
 """
-
+import collections
 import os
 import errno
 import argparse
@@ -52,6 +52,7 @@ def check_samplesheet(file_in, file_out):
     """
 
     sample_mapping_dict = {}
+    input_dict = collections.defaultdict(list)
     with open(file_in, "r", encoding="utf-8-sig") as fin:
         ## Check header
         MIN_COLS = 2
@@ -144,6 +145,7 @@ def check_samplesheet(file_in, file_out):
                         "Line",
                         line,
                     )
+            is_control_input = not antibody and not control
 
             ## Auto-detect paired-end/single-end
             if not sample or not fastq_1:
@@ -172,7 +174,9 @@ def check_samplesheet(file_in, file_out):
                     print_error("Samplesheet contains duplicate rows!", "Line", line)
                 else:
                     sample_mapping_dict[sample].append(sample_info)
-    # pprint.pprint(sample_mapping_dict)
+            if is_control_input:
+                input_dict[sample_basename].append(sample_info)
+
     ## Write validated samplesheet with appropriate columns
     if len(sample_mapping_dict) > 0:
         out_dir = os.path.dirname(file_out)
@@ -205,11 +209,12 @@ def check_samplesheet(file_in, file_out):
                         sample,
                     )
 
+                # check that the control/input exists
                 for idx, val in enumerate(sample_mapping_dict[sample]):
                     control = val[-1]
-                    if control and control not in sample_mapping_dict.keys():
+                    if control and control not in input_dict.keys():
                         print_error(
-                            f"Control identifier has to match a provided sample identifier!",
+                            "Control identifier has to match a provided sample identifier!",
                             "Control",
                             control,
                         )

main.nf

@@ -21,6 +21,7 @@ log.info """\
 include { FASTQ_DOWNLOAD_PREFETCH_FASTERQDUMP_SRATOOLS as DOWNLOAD_FASTQ } from './subworkflows/nf-core/fastq_download_prefetch_fasterqdump_sratools'
 include { INPUT_CHECK              } from './subworkflows/local/input_check.nf'
 include { PREPARE_GENOME           } from './subworkflows/local/prepare_genome.nf'
+include { POOL_INPUTS              } from './subworkflows/local/pool_inputs/'
 include { FILTER_BLACKLIST         } from './subworkflows/CCBR/filter_blacklist/'
 include { ALIGN_GENOME             } from "./subworkflows/local/align.nf"
 include { DEDUPLICATE              } from "./subworkflows/local/deduplicate.nf"
@@ -74,28 +75,27 @@ workflow CHIPSEQ {
     INPUT_CHECK(file(params.input, checkIfExists: true), params.seq_center, contrast_sheet)
 
     INPUT_CHECK.out.reads.set { raw_fastqs }
-    raw_fastqs | CUTADAPT
-    CUTADAPT.out.reads.set{ trimmed_fastqs }
+    CUTADAPT(raw_fastqs).reads | POOL_INPUTS
+    trimmed_fastqs = POOL_INPUTS.out.reads
 
     PREPARE_GENOME()
     chrom_sizes = PREPARE_GENOME.out.chrom_sizes
-
     effective_genome_size = PREPARE_GENOME.out.effective_genome_size
+
     FILTER_BLACKLIST(trimmed_fastqs, PREPARE_GENOME.out.blacklist_index)
     ALIGN_GENOME(FILTER_BLACKLIST.out.reads, PREPARE_GENOME.out.reference_index)
-    ALIGN_GENOME.out.bam.set{ aligned_bam }
+    aligned_bam = ALIGN_GENOME.out.bam
 
     DEDUPLICATE(aligned_bam, chrom_sizes, effective_genome_size)
-    DEDUPLICATE.out.bam.set{ deduped_bam }
-    DEDUPLICATE.out.tag_align.set{ deduped_tagalign }
+    deduped_bam = DEDUPLICATE.out.bam
+    deduped_tagalign = DEDUPLICATE.out.tag_align
 
-    deduped_bam | PHANTOM_PEAKS
-    PHANTOM_PEAKS.out.fraglen | PPQT_PROCESS
-    PPQT_PROCESS.out.fraglen.set { frag_lengths }
+    PHANTOM_PEAKS(deduped_bam).fraglen | PPQT_PROCESS
+    frag_lengths = PPQT_PROCESS.out.fraglen
 
     ch_multiqc = Channel.of()
     if (params.run.qc) {
-        QC(raw_fastqs, trimmed_fastqs, FILTER_BLACKLIST.out.n_surviving_reads,
+        QC(raw_fastqs, CUTADAPT.out.reads, FILTER_BLACKLIST.out.n_surviving_reads,
            aligned_bam, ALIGN_GENOME.out.aligned_flagstat, ALIGN_GENOME.out.filtered_flagstat,
            deduped_bam, DEDUPLICATE.out.flagstat,
            PHANTOM_PEAKS.out.spp, frag_lengths,
@@ -157,6 +157,7 @@ workflow CHIPSEQ {
                 )
 
         }
+
     }
 
     if (!workflow.stubRun) {

modules.json

@@ -100,6 +100,11 @@
                         "git_sha": "8fc1d24c710ebe1d5de0f2447ec9439fd3d9d66a",
                         "installed_by": ["modules"]
                     },
+                    "cat/cat": {
+                        "branch": "master",
+                        "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
+                        "installed_by": ["modules"]
+                    },
                     "custom/sratoolsncbisettings": {
                         "branch": "master",
                         "git_sha": "20e78a9868eaa69c8cac91152397def32374b807",