Merge pull request #146 from emreerhan/abyss

Added an abyss component to flowcraft, and added a new disableRR parameter to spades

Author: ODiogoSilva

changelog.md

@@ -2,6 +2,15 @@
 
 ## Changes in upcoming release (`dev` branch) 
 
+### Components changes
+
+- Added new `disableRR` param in the `spades` component that disables repeat
+resolution
+
+### New components
+
+- Added component `abyss`.
+
 ### Minor/Other changes
 
 - Added removal of duplicate IDs from `reads_download` component input.
@@ -27,6 +36,7 @@ which is particularly useful in very large workflows.
 `mapping_patlas`.
 
 ### New components
+
 - Added component `fast_ani`.
 
 ### Minor/Other changes

flowcraft/generator/components/assembly.py

@@ -58,6 +58,11 @@ def __init__(self, **kwargs):
                     "is only useful to remove temporary files in large "
                     "workflows and prevents nextflow's resume functionality. "
                     "Use with caution."
+            },
+            "disableRR": {
+                "default": "false",
+                "description":
+                    "disables repeat resolution stage of assembling."
             }
         }
 
@@ -177,3 +182,37 @@ def __init__(self, **kwargs):
                     "Use with caution."
             }
         }
+
+class Abyss(Process):
+    """ABySS process template interface
+
+    This process is set with:
+
+        - ``input_type``: fastq
+        - ``output_type``: assembly
+        - ``ptype``: assembly
+
+    """
+
+    def __init__(self, **kwargs):
+
+        super().__init__(**kwargs)
+
+        self.input_type = "fastq"
+        self.output_type = "fasta"
+
+        self.params = {
+            "abyssKmer": {
+                "default": "96",
+                "description":
+                    "kmer size for assembly."
+            }
+        }
+
+        self.directives = {"abyss": {
+            "cpus": 4,
+            "memory": "{ 5.GB * task.attempt }",
+            "container": "flowcraft/abyss",
+            "version": "2.1.1",
+            "scratch": "true"
+        }}

flowcraft/generator/engine.py

@@ -55,6 +55,7 @@
 
 
 process_map = {
+        "abyss": assembly.Abyss,
         "abricate": annotation.Abricate,
         "assembly_mapping": ap.AssemblyMapping,
         "bowtie": mapping.Bowtie,
@@ -101,7 +102,8 @@
         "split_assembly": meta.SplitAssembly,
         "trimmomatic": readsqc.Trimmomatic,
         "true_coverage": readsqc.TrueCoverage,
-        "viral_assembly": assembly.ViralAssembly
+        "viral_assembly": assembly.ViralAssembly,
+        "abyss": assembly.Abyss
 }
 
 """

flowcraft/generator/templates/abyss.nf

@@ -0,0 +1,26 @@
+
+IN_abyss_kmer_{{ pid }} = Channel.value(params.abyssKmer{{ param_id }})
+
+process abyss_{{ pid }} {
+    
+    // Send POST request to platform
+    {% include "post.txt" ignore missing %}
+
+    tag { sample_id }
+    publishDir 'results/assembly/abyss_{{ pid }}/', pattern: '*-scaffolds.fa', mode: 'copy'
+
+    input:
+    set sample_id, file(fastq_pair) from {{ input_channel }}
+    val kmer from IN_abyss_kmer_{{ pid }}
+
+    output:
+    set sample_id, file('*.fa') into {{ output_channel }}
+    {% with task_name="abyss" %}
+    {%- include "compiler_channels.txt" ignore missing -%}
+    {% endwith %}
+
+    script:
+    "abyss-pe in=\"${fastq_pair[0]} ${fastq_pair[1]}\" k=${kmer} name=${sample_id}"
+}
+
+{{ forks }}

flowcraft/generator/templates/spades.nf

@@ -5,7 +5,10 @@ if ( !params.spadesMinKmerCoverage{{ param_id }}.toString().isNumber()){
     exit 1, "'spadesMinKmerCoverage{{ param_id }}' parameter must be a number. Provided value: '${params.spadesMinKmerCoverage{{ param_id }}}'"
 }
 
-IN_spades_opts_{{ pid }} = Channel.value([params.spadesMinCoverage{{ param_id }},params.spadesMinKmerCoverage{{ param_id }}])
+IN_spades_opts_{{ pid }} = Channel.value(
+    [params.spadesMinCoverage{{ param_id }},
+     params.spadesMinKmerCoverage{{ param_id }}
+     ])
 
 if ( params.spadesKmers{{ param_id }}.toString().split(" ").size() <= 1 ){
     if (params.spadesKmers{{ param_id }}.toString() != 'auto'){
@@ -15,7 +18,10 @@ if ( params.spadesKmers{{ param_id }}.toString().split(" ").size() <= 1 ){
 IN_spades_kmers_{{pid}} = Channel.value(params.spadesKmers{{ param_id }})
 
 clear = params.clearInput{{ param_id }} ? "true" : "false"
+disable_rr = params.disableRR{{ param_id }} ? "true" : "false"
+
 checkpointClear_{{ pid }} = Channel.value(clear)
+disableRR_{{ pid }} = Channel.value(disable_rr)
 
 process spades_{{ pid }} {
 
@@ -32,6 +38,7 @@ process spades_{{ pid }} {
     val opts from IN_spades_opts_{{ pid }}
     val kmers from IN_spades_kmers_{{ pid }}
     val clear from checkpointClear_{{ pid }}
+    val disable_rr from disableRR_{{ pid }}
 
     output:
     set sample_id, file('*_spades*.fasta') into {{ output_channel }}

flowcraft/templates/spades.py

@@ -85,6 +85,7 @@ def __get_version_spades():
     MAX_LEN = int('$max_len'.strip())
     KMERS = '$kmers'.strip()
     CLEAR = '$clear'
+    DISABLE_RR = '$disable_rr'
     OPTS = [x.strip() for x in '$opts'.strip("[]").split(",")]
     CLEAR = '$clear'
     logger.debug("Running {} with parameters:".format(
@@ -95,6 +96,7 @@ def __get_version_spades():
     logger.debug("KMERS: {}".format(KMERS))
     logger.debug("OPTS: {}".format(OPTS))
     logger.debug("CLEAR: {}".format(CLEAR))
+    logger.debug("DISABLE_RR: {}".format(DISABLE_RR))
 
 
 def set_kmers(kmer_opt, max_read_len):
@@ -163,7 +165,7 @@ def clean_up(fastq):
 
 
 @MainWrapper
-def main(sample_id, fastq_pair, max_len, kmer, opts, clear):
+def main(sample_id, fastq_pair, max_len, kmer, opts, clear, disable_rr):
     """Main executor of the spades template.
 
     Parameters
@@ -182,7 +184,9 @@ def main(sample_id, fastq_pair, max_len, kmer, opts, clear):
     clear : str
         Can be either 'true' or 'false'. If 'true', the input fastq files will
         be removed at the end of the run, IF they are in the working directory
-
+    disable_rr : str
+        Can either be 'true' or 'false'. If 'true', disables repeat resolution 
+        stage of assembling
     """
 
     logger.info("Starting spades")
@@ -217,6 +221,10 @@ def main(sample_id, fastq_pair, max_len, kmer, opts, clear):
         fastq_pair[1]
     ]
 
+    # Disable RR?
+    if disable_rr == 'true':
+        cli += ['--disable-rr']
+
     logger.debug("Running SPAdes subprocess with command: {}".format(cli))
 
     p = subprocess.Popen(cli, stdout=PIPE, stderr=PIPE)
@@ -263,5 +271,4 @@ def main(sample_id, fastq_pair, max_len, kmer, opts, clear):
 
 
 if __name__ == '__main__':
-
-    main(SAMPLE_ID, FASTQ_PAIR, MAX_LEN, KMERS, OPTS, CLEAR)
+    main(SAMPLE_ID, FASTQ_PAIR, MAX_LEN, KMERS, OPTS, CLEAR, DISABLE_RR)