unicycler: Add parameter --long_reads

changelog.md

@@ -16,11 +16,13 @@ resolution
 - The `abyss` and `spades` components emit GFA in a secondary channel.
 - The new `bandage` component can accept either FASTA from a primary channel
   or GFA from a secondary channel.
+- The new `unicycler` component can accept long reads from a secondary channel.
 
 ### New components
 
 - Added component `abyss`.
 - Added component `bandage`.
+- Added component `porechop`.
 - Added component `unicycler`.
 
 ### Minor/Other changes

flowcraft/generator/components/assembly.py

@@ -236,8 +236,17 @@ def __init__(self, **kwargs):
 
         self.input_type = "fastq"
         self.output_type = "fasta"
+
+        self.link_end.append({"link": "long_reads", "alias": "long_reads"})
         self.link_start.append("gfa1")
 
+        self.params = {
+            "long_reads": {
+                "default": "null",
+                "description": "FASTQ or FASTA file of long reads"
+            },
+        }
+
         self.directives = {"unicycler": {
             "cpus": 4,
             "container": "quay.io/biocontainers/unicycler",

flowcraft/generator/components/reads_quality_control.py

@@ -433,3 +433,37 @@ def __init__(self, **kwargs):
         self.status_channels = [
             "downsample_fastq"
         ]
+
+class Porechop(Process):
+    """Porechop trims adapters from Oxford Nanopore reads.
+
+    This process is set with:
+
+        - ``input_type``: fastq
+        - ``output_type``: fastq
+        - ``ptype``: pre_assembly
+    """
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+        self.input_type = "fastq"
+        self.output_type = "fastq"
+
+        self.link_end.append({"link": "raw_long_reads", "alias": "raw_long_reads"})
+        self.link_start.append("long_reads")
+
+        self.params = {
+            "long_reads": {
+                "default": "null",
+                "description": "FASTQ or FASTA file of long reads"
+            },
+        }
+
+        self.directives = {
+            "porechop": {
+                "cpus": 4,
+                "container": "quay.io/biocontainers/porechop",
+                "version": "0.2.3_seqan2.1.1--py36h2d50403_3"
+            }
+        }

flowcraft/generator/engine.py

@@ -88,6 +88,7 @@
         "momps": typing.Momps,
         "patho_typing": typing.PathoTyping,
         "pilon": ap.Pilon,
+        "porechop": readsqc.Porechop,
         "process_skesa": ap.ProcessSkesa,
         "process_spades": ap.ProcessSpades,
         "progressive_mauve":alignment.ProgressiveMauve,

flowcraft/generator/templates/porechop.nf

@@ -0,0 +1,28 @@
+// True when a raw_long_reads secondary channel is connected to this component.
+has_raw_long_reads_{{pid}} = binding.hasVariable('raw_long_reads_{{pid}}')
+
+process porechop_{{pid}} {
+    {% include "post.txt" ignore missing %}
+
+    publishDir "results/porechop_{{pid}}", pattern: "*.fastq.gz"
+    publishDir "reports/porechop_{{pid}}", pattern: "*.log"
+
+    tag { sample_id }
+
+    input:
+    set sample_id, file(fastq_pair) from {{input_channel}}
+    file raw_long_reads from has_raw_long_reads_{{pid}} ? raw_long_reads_{{pid}} :
+        Channel.fromPath(params.long_reads{{param_id}})
+
+    output:
+    set sample_id, file(fastq_pair) into {{output_channel}}
+    file "${sample_id}.porechop.fastq.gz" into long_reads_{{pid}}
+    {% with task_name="porechop" %}
+    {%- include "compiler_channels.txt" ignore missing -%}
+    {% endwith %}
+
+    script:
+    "time porechop -t $task.cpus --format fastq.gz -i ${raw_long_reads} -o ${sample_id}.porechop.fastq.gz >${sample_id}.log"
+}
+
+{{ forks }}

flowcraft/generator/templates/unicycler.nf

@@ -1,3 +1,6 @@
+// True when a long_reads secondary channel is connected to this component.
+has_long_reads_{{pid}} = binding.hasVariable('long_reads_{{pid}}')
+
 process unicycler_{{pid}} {
     {% include "post.txt" ignore missing %}
 
@@ -7,6 +10,9 @@ process unicycler_{{pid}} {
 
     input:
     set sample_id, file(fastq_pair) from {{input_channel}}
+    file long_reads from has_long_reads_{{pid}} ? long_reads_{{pid}} :
+        params.long_reads{{param_id}} ? Channel.fromPath(params.long_reads{{param_id}}) :
+        Channel.value("NA")
 
     output:
     set sample_id, file('assembly.fasta') into {{output_channel}}
@@ -16,7 +22,10 @@ process unicycler_{{pid}} {
     {% endwith %}
 
     script:
-    "unicycler -t $task.cpus -o . --no_correct --no_pilon -1 ${fastq_pair[0]} -2 ${fastq_pair[1]}"
+    command = "unicycler -t $task.cpus -o . --no_correct --no_pilon -1 ${fastq_pair[0]} -2 ${fastq_pair[1]}"
+    if (params.long_reads{{param_id}})
+        command += " -l ${long_reads}"
+    command
 }
 
 {{forks}}