feat: expanding support for drep and bin consolidation

config/cluster.json

@@ -9,7 +9,8 @@
         "threads": 8,
         "mem": "16g",
         "partition": "norm",
-        "time": "0-04:00:00"
+        "time": "0-04:00:00",
+        "gres": "None"
     },
     "metawrap_genome_assembly": {
         "threads": 24,

config/images.json

@@ -1,6 +1,10 @@
 {
     "images": {
-        "metawrap": "docker://rroutsong/metamorph_metawrap:1.3.2",
+        "metawrap": "docker://rroutsong/metamorph_metawrap:0.0.2",
         "metagenome": "docker://rroutsong/metamorph_metagenome:0.0.1"
+    }, 
+    "containers": {
+        "metawrap": "/data/OpenOmics/SIFs/metamorph_metawrap_1.3.2.sif",
+        "metagenome": "/data/OpenOmics/SIFs/metamorph_metagenome_0.0.1.sif"
     }
 }

metamorph

@@ -46,16 +46,15 @@ import argparse  # potential python3 3rd party package, added in python/3.5
 # Local imports
 from src import version 
 from src.run import init, setup, bind, dryrun, runner, run_coa_pipeline
-from src.shells import bash 
 from src.utils import (
     Colors,
     err,
     exists,
     fatal,
-    hashed,
     permissions,
     check_cache,
-    require)
+    require
+)
 
 
 # Pipeline Metadata

src/run.py

@@ -4,6 +4,8 @@
 # Python standard library
 from __future__ import print_function
 from shutil import copytree
+from uuid import uuid4
+from datetime import datetime
 import os, re, json, sys, subprocess
 
 # Local imports
@@ -329,11 +331,9 @@ def bind(sub_args, config):
             if value not in bindpaths:
                 bindpaths.append(value)
 
-    rawdata_bind_paths = [os.path.realpath(p) for p in config['project']['datapath'].split(',')]
+    rawdata_bind_paths = [os.path.abspath(p) for p in config['project']['datapath'].split(',')]
     working_directory =  os.path.realpath(config['project']['workpath'])
 
-
-
     return bindpaths
 
 
@@ -393,6 +393,8 @@ def add_user_information(config):
     # username
     config['project']['userhome'] = home
     config['project']['username'] = username
+    dt = datetime.now().strftime("%m/%d/%Y")
+    config['project']['id'] = f"{uuid4()}_{dt}_metagenome"
 
     return config
 

workflow/Snakefile

@@ -24,42 +24,49 @@ configfile: "config.json"  # Generated from user input and config/*.json
 rule all:
     input:
         # read qc and filtering
-        # ~~~~~~~~~~~~~~~~~~~~~
+        # ~~~~~~~~~~~~~~~
         expand(
             join(
-                workpath, "metawrap_readqc", "{name}", "{name}.R1_bmtagger_report.html"
+                workpath, config['project']['id'], "metawrap_qc", "{name}", "{name}.R1_bmtagger_report.html"
             ),
             name=samples,
         ),
         expand(
             join(
-                workpath, "metawrap_readqc", "{name}", "{name}.R2_bmtagger_report.html"
+                workpath, config['project']['id'], "metawrap_qc", "{name}", "{name}.R2_bmtagger_report.html"
             ),
             name=samples,
         ),
-        expand(join(workpath, "{name}", "{name}.R1_readqc.fastq"), name=samples),
-        expand(join(workpath, "{name}", "{name}.R2_readqc.fastq"), name=samples),
         # genome assembly
         # ~~~~~~~~~~~~~~~
         expand(
-            join(workpath, "metawrap_assembly", "{name}", "final_assembly.fasta"),
+            join(workpath, config['project']['id'], "metawrap_assembly", "{name}", "final_assembly.fasta"),
             name=samples,
         ),
         expand(
-            join(workpath, "metawrap_assembly", "{name}", "assembly_report.html"),
+            join(workpath, config['project']['id'], "metawrap_assembly", "{name}", "assembly_report.html"),
             name=samples,
         ),
         # taxonomic classification
-        # ~~~~~~~~~~~~~~~~~~~~~~~~
-        expand(join(workpath, "metawrap_taxonomy", "{name}", "final_assembly.krona"), name=samples),
-        expand(join(workpath, "metawrap_taxonomy", "{name}", "kronagram.html"), name=samples),
-        expand(join(workpath, "metawrap_taxonomy", "{name}", "final_assembly.krak2"), name=samples),
-        expand(join(workpath, "metawrap_taxonomy", "{name}", "final_assembly.kraken2"), name=samples),
-        # assembly binning and refinement
-        # ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-        expand(join(workpath, "metawrap_binning", "{name}"), name=samples),
-        expand(join(workpath, "metawrap_refine_bins", "{name}"),  name=samples),
+        # ~~~~~~~~~~~~~~~
+        expand(join(workpath, config['project']['id'], "metawrap_kmer", "{name}", "final_assembly.krona"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_kmer", "{name}", "kronagram.html"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_kmer", "{name}", "final_assembly.krak2"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_kmer", "{name}", "final_assembly.kraken2"), name=samples),
+        # assembly binning
+        # ~~~~~~~~~~~~~~~
+        expand(join(workpath, config['project']['id'], "metawrap_binning", "{name}", "maxbin2_bins.contigs"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_binning", "{name}", "maxbin2_bins.stats"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_binning", "{name}", "metabat2_bins.contigs"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_binning", "{name}", "metabat2_bins.stats"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_binning", "{name}", "metawrap_50_5_bins.contigs"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_binning", "{name}", "metawrap_50_5_bins.stats"), name=samples),
+        expand(join(workpath, config['project']['id'], "metawrap_binning", "{name}", "figures", "binning_results.png"), name=samples),
+        # bin refinement
+        # ~~~~~~~~~~~~~~~
+        expand(join(workpath, config['project']['id'], "metawrap_bin_refine", "{name}", "dereplicated_bins"), name=samples),
+
 
 
 # Import rules
-include: join("rules", "metawrap.smk")
+include: join("rules", "metawrap.smk")