Merge branch 'Alia' into 'master'

Alia

See merge request testoni-lab/bolero!8

src/.docker_modules/pycoqc/2.5.2/Dockerfile

-FROM quay.io/biocontainers/pycoqc:2.5.2--py_0
\ No newline at end of file
+FROM python:3.10
+
+# copy the dependencies file to the working directory
+COPY requirements.txt .
+
+# install dependencies
+RUN pip install --upgrade pip setuptools wheel
+RUN pip install -r requirements.txt
+
+# install pycoQC from pypi
+RUN pip install pycoQC
+
+# command to run on container start
+CMD [ "bash" ]
\ No newline at end of file

src/.docker_modules/pycoqc/2.5.2/requirements.txt

+numpy>=1.13
+scipy>=1.1
+pandas>=0.23
+plotly>=3.4
+jinja2>=2.10
+h5py>=2.8.0
+tqdm>=4.23

src/bolero.nf

@@ -45,6 +45,10 @@ def helpMessage() {
       --gpu_mode [boolean]            Guppy basecaller configuration. Default: false.
                                       "gpu" mode is dedicated to NVIDIA Cuda compatible system according to Guppy specifications.
     
+    Nanopore barcoding:
+      --kit_barcoding                 Nanopore barcoding kit.
+      --config_file                   Nanopore configuration file.
+    
     GPU basecalling parameters:
       --min_qscore [float]            Minimum quality score threshold, default = 7.0.
       --gpu_runners_per_device [int]  Number of runner per device, default = 32 (refer to guppy manual).
@@ -93,7 +97,6 @@ params.chunks_per_runner = 512
 params.chunk_size = 1900
 params.config_file = ""
 params.kit_barcoding = ""
-//"EXP-PBC001"
 
 
 /* Params out */
@@ -133,18 +136,6 @@ Channel
     .ifEmpty { error "No fast5/q folder defined." }
     .set { input }
 
-/*
-Channel
-    .fromPath( params.adapt )
-    .ifEmpty { error "No adapter sequence defined." }
-    .set { adapt }
-
-Channel
-    .fromPath( params.gsp )
-    .ifEmpty { error "No adapter sequence defined." }
-    .set { gsp }
-*/
-
 Channel
     .fromPath( params.genome )
     .ifEmpty { error "No genome defined, a fasta file containing the full length preC RNA from HBV genome." }
@@ -161,13 +152,6 @@ Channel
     .set{barcodes}
 
 
-// Test pour barcoding process
-Channel
-    .fromPath('/home/alia/pipelines/bolero/results/01_Basecalling/pass/', type: 'dir')
-    .set{pass}
-Channel
-    .fromPath('/home/alia/pipelines/bolero/results/01_Basecalling/sequencing_summary.txt')
-    .set{ss}
 /*
  ****************************************************************
                           Imports
@@ -211,18 +195,11 @@ include { rna_count } from "./nf_modules/rna_count/main.nf"
 */
 
 workflow {
-  /*
-  pass.view()
-  barcoding_cpu(pass)
-  */
-  //######################## BASECALLING ########################
-
 
   if(params.skipBC) { // we take fastq files as input and skip basecalling
     concatenate(barcodes)
   }
 
-  //il reste à adapter ça
   else { // we take fast5 files as input and proceed to basecalling with guppy
     if(params.gpu_mode) {
       basecall_fast5_gpu(input)
@@ -240,13 +217,12 @@ workflow {
           .set{tuple_sample}
         concatenate(tuple_sample)
       }
-      control_basecalling(basecall_fast5_gpu.out.sequencing_summary)
 
     }
     else {
-      //basecall_fast5_cpu(input)
+      basecall_fast5_cpu(input)
       if(params.kit_barcoding != ""){
-        barcoding_cpu(pass)  
+        barcoding_cpu(basecall_fast5_cpu.out.pass)  
         barcoding_cpu.out.barcodes
           .flatten()
           .map{it -> [it.name, it]}
@@ -254,12 +230,11 @@ workflow {
         concatenate(tuples_barcode)
       }
       else{
-        pass
+        basecall_fast5_cpu.out.pass
           .map{it -> ["Sample", it]}
           .set{tuple_sample}
         concatenate(tuple_sample)
       }
-      control_basecalling(ss)
     }
   }
 
@@ -277,23 +252,37 @@ workflow {
   //########################## MAPPING ##########################
 
   hbv_genome(cut_5pRACE.out.fastq_cutadapt, genome.collect())
-  sort_index_bam(hbv_genome.out.bam)
-  //il faut ajouter une boucle if pour le mode cpu/gpu
-  control_bam(basecall_fast5_gpu.out.sequencing_summary.collect(), sort_index_bam.out.indexed_bam)
 
+  //Index
+  sort_index_bam(hbv_genome.out.bam)
+  
+  //Quality control
+  if(params.skipBC == False) {
+    if(params.gpu_mode) {
+      control_bam(basecall_fast5_gpu.out.sequencing_summary.collect(), sort_index_bam.out.indexed_bam)
+    }
+    else {
+      control_bam(basecall_fast5_cpu.out.sequencing_summary.collect(), sort_index_bam.out.indexed_bam)
+    }
+  }
   //###################### START POSITIONS #######################
 
+  //Identification of start positions
   start_position_counts(sort_index_bam.out.indexed_bam)
+
+  //Identification of RNA
   start_position_individuals(start_position_counts.out.count)
 
   //#################### VARIANTS D'EPISSAGE ####################
 
+  //Identification of splicing junction sites
   jwr_checker(sort_index_bam.out.indexed_bam)
 
   start_position_individuals.out.classification_of_reads
     .combine(jwr_checker.out.nanosplicer_jwr, by: 0)
     .set{files_for_nanosplicer}
 
+  //Identification of variants
   junctions_nanosplicer(files_for_nanosplicer)
   
   //#################### VARIANTS D'EPISSAGE ####################
@@ -302,6 +291,7 @@ workflow {
     .combine(start_position_individuals.out.classification_of_reads, by: 0)
     .set{files_for_rna_count}
 
+  //Variants count
   rna_count(files_for_rna_count)
 
 }

src/nf_modules/ont-guppy/main.nf

@@ -60,8 +60,7 @@ guppy_basecaller --compress_fastq \
    --gpu_runners_per_device ${params.gpu_runners_per_device} \
    --num_callers ${params.num_callers} \
    --chunks_per_runner ${params.chunks_per_runner} \
-   --flowcell ${params.flowcell} \
-   --kit ${params.kit}
+   ${options}
 """
 }
 
@@ -129,6 +128,13 @@ process barcoding_gpu {
     publishDir "results/${params.barcoding_out}", mode: 'copy'
   }
 
+  if (params.config_file != "") {
+    options = "-c /opt/ont/guppy/data/${params.config_file} \ "
+  }
+  else {
+    options = ""
+  }
+
   input:
     path(pass_path)
   
@@ -142,6 +148,7 @@ guppy_barcoder \
   --save_path . \
   -t ${params.gpu_threads_per_caller} \
   --barcode_kits ${params.kit_barcoding} \
+  ${options}
   --progress_stats_frequency 60 \
   --enable_trim_barcodes \
   --trim_adapters \
@@ -157,6 +164,13 @@ process barcoding_cpu {
     publishDir "results/${params.barcoding_out}", mode: 'copy'
   }
 
+  if (params.config_file != "") {
+    options = "-c /opt/ont/guppy/data/${params.config_file} \ "
+  }
+  else {
+    options = ""
+  }
+
   input:
     path(pass_path)
   
@@ -171,6 +185,7 @@ guppy_barcoder \
   --save_path . \
   -t ${task.cpus} \
   --barcode_kits ${params.kit_barcoding} \
+  ${options}
   --progress_stats_frequency 60 \
   --enable_trim_barcodes \
   --trim_adapters \