Add rna inputs workflow (#11)

* feat: accept inputs for both dna and rna
OpenOmics · Feb 28, 2024 · e5fc0b9 · e5fc0b9
1 parent db7f9de
commit e5fc0b9
Show file tree

Hide file tree

Showing 20 changed files with 976 additions and 689 deletions.
diff --git a/config/cluster.json b/config/cluster.json
@@ -1,32 +1,60 @@
 {
     "__default__": {
         "threads": 4,
-        "mem": "8g",
+        "mem": "8G",
         "partition": "norm",
-        "time": "0-04:00:00"
+        "time": "0-04:00:00",
+        "gres": "lscratch:64"
+    },
+    "concat_reads": {
+        "threads": 28,
+        "mem": "16G",
+        "partition": "norm",
+        "time": "1-00:00:00",
+        "gres": "lscratch:400"
+    },
+    "concat_rna_reads": {
+        "threads": 28,
+        "mem": "16G",
+        "partition": "norm",
+        "time": "1-00:00:00",
+        "gres": "lscratch:400"
     },
     "metawrap_read_qc": {
-        "threads": 8,
-        "mem": "16g",
+        "threads": 16,
+        "mem": "32G",
         "partition": "norm",
-        "time": "0-04:00:00",
-        "gres": "None"
+        "time": "1-00:00:00",
+        "gres": "lscratch:400"
+    },
+    "rna_read_qc": {
+        "threads": 16,
+        "mem": "32G",
+        "partition": "norm",
+        "time": "1-00:00:00",
+        "gres": "lscratch:400"
     },
     "metawrap_genome_assembly": {
-        "threads": 24,
-        "mem": "128g",
+        "threads": 48,
+        "mem": "128G",
         "partition": "norm",
-        "time": "2-00:00:00"
+        "time": "5-00:00:00"
     },
     "metawrap_tax_classification": {
-        "threads": 12,
-        "mem": "32g",
-        "partition": "quick",
-        "time": "0-04:00:00"
-    },
-    "metawrap_assembly_binning": {
-        "threads": 16,
-        "mem": "64g",
+        "threads": 32,
+        "mem": "64G",
+        "partition": "norm",
+        "time": "5-00:00:00"
+    }, 
+    "metawrap_binning": {
+        "threads": 32,
+        "mem": "64G",
+        "partition": "norm",
+        "time": "2-00:00:00"
+    }, 
+    "derep_bins": {
+        "threads": 32,
+        "mem": "32G",
         "partition": "norm",
         "time": "2-00:00:00"
     }

diff --git a/config/images.json b/config/images.json
@@ -1,10 +1,10 @@
 {
     "images": {
-        "metawrap": "docker://rroutsong/metamorph_metawrap:0.0.2",
+        "metawrap": "docker://rroutsong/metamorph_metawrap:0.0.4",
         "metagenome": "docker://rroutsong/metamorph_metagenome:0.0.1"
     }, 
     "containers": {
-        "metawrap": "/data/OpenOmics/SIFs/metamorph_metawrap_1.3.2.sif",
+        "metawrap": "/data/OpenOmics/SIFs/metamorph_metawrap_0.0.4.sif",
         "metagenome": "/data/OpenOmics/SIFs/metamorph_metagenome_0.0.1.sif"
     }
 }
diff --git a/config/resources.json b/config/resources.json
@@ -1,36 +1,46 @@
 {
-    "binds": [
+    "databases": [
         {
-            "name": "KRAKEN_DB2",
-            "to": "$HOME/KRAKEN_DB2",
-            "from": "/data/OpenOmics/references/metamorph/kraken2/k2_pluspfp_08gb_20230605",
-            "mode": "ro"
+            "name": "KRAKEN2_DB",
+            "to": "/data2/KRAKEN_DB2",
+            "from": "/data/OpenOmics/references/metamorph/kraken2/k2_pluspfp_08gb_20240112",
+            "mode": "rw"
         }, {
             "name": "KRAKEN_DB",
-            "to": "$HOME/KRAKEN_DB",
+            "to": "/data2/KRAKEN_DB",
             "from": "/data/OpenOmics/references/metamorph/kraken/20171019_minikraken_8GB",
             "mode": "ro"
         }, {
             "name": "BMTAGGER_INDEX",
-            "to": "$HOME/BMTAGGER_DB",
+            "to": "/data2/BMTAGGER_DB",
             "from": "/data/OpenOmics/references/metamorph/BMTAGGER/BMTAGGER_INDEX",
             "mode": "ro"
         }, {
             "name": "BMTAGGER_INDEX",
-            "to": "$HOME/GTDBTK_DB",
+            "to": "/data2/GTDBTK_DB",
             "from": "/data/OpenOmics/references/metamorph/GTDBtk/release207_v2",
             "mode": "ro"
         }, {
             "name": "GUNC_DB",
-            "to": "$HOME/GUNC_DB",
+            "to": "/data2/GUNC_DB",
             "from": "/data/OpenOmics/references/metamorph/GUNC/gunc_1.0.5db",
             "mode": "ro"
+        }, {
+            "name": "NCBI_NT",
+            "to": "/data2/NCBI_NT_DB",
+            "from": "/data/OpenOmics/references/metamorph/NCBI_nt",
+            "mode": "rw"
+        }, {
+            "name": "NCBI_TAX",
+            "to": "/data2/NCBI_TAX_DB",
+            "from": "/data/OpenOmics/references/metamorph/taxonomy",
+            "mode": "rw"
         }, {
             "name": "CHECKM_DB",
-            "to": "$HOME/checkm",
+            "to": "/data2/CHECKM_DB",
             "from": "/data/OpenOmics/references/metamorph/checkm",
             "mode": "rw"
-        }, {
+        },  {
             "name": "CHECKM_CONFIG",
             "to": "/opt/conda/envs/metawrap-env/lib/python2.7/site-packages/checkm/DATA_CONFIG",
             "from": "/data/OpenOmics/references/metamorph/checkm/DATA_CONFIG",

diff --git a/docker/Dockerfile b/docker/Dockerfile
diff --git a/docker/metawrap/Dockerfile b/docker/metawrap/Dockerfile
@@ -0,0 +1,23 @@
+FROM condaforge/miniforge3:latest
+RUN apt-get update; apt-get install -y -qq curl build-essential vim dos2unix bash python2.7
+RUN mamba create -y -n metawrap-env
+RUN conda config --add channels defaults; conda config --add channels conda-forge; \
+    conda config --add channels bioconda;conda config --add channels ursky
+RUN mamba install -y --only-deps -c ursky -n metawrap-env metawrap-mg==1.3.2
+RUN cd /home; git clone https://github.com/rroutsong/metaWRAP.git; chmod -R 777 metaWRAP
+RUN mamba run -n metawrap-env pip3 install drep
+ENV PATH="/home/metaWRAP/bin:/opt/conda/envs/metawrap-env/bin:$PATH"
+COPY docker/metawrap/config-metawrap /home/metaWRAP/bin/config-metawrap
+COPY docker/metawrap/Dockerfile /Dockerfile
+RUN mkdir /install; cd /install; wget https://carlowood.github.io/which/which-2.21.tar.gz; tar xvf which-2.21.tar.gz
+RUN cd /install/which-2.21; ./configure; make && make install
+RUN rm /usr/bin/which; ln -s /usr/local/bin/which /usr/bin/which
+RUN md5sum Dockerfile > /Dockerfile.md5
+ADD docker/metawrap/mw /home/metaWRAP/bin/mw
+RUN dos2unix /home/metaWRAP/bin/config-metawrap
+RUN dos2unix /home/metaWRAP/bin/mw
+RUN chmod a+rx /home/metaWRAP/bin/config-metawrap /home/metaWRAP/bin/mw
+ENV BASH_ENV="/etc/bash.bashrc"
+RUN echo ". /opt/conda/etc/profile.d/conda.sh && conda activate metawrap-env" >> /etc/bash.bashrc
+RUN echo 'export PATH="/home/metaWRAP/bin:/opt/conda/envs/metawrap-env/bin:$PATH"' >> /etc/bash.bashrc
+ENTRYPOINT ["/bin/bash"]
diff --git a/docker/metawrap/config-metawrap_1.3.2 → docker/metawrap/config-metawrap b/docker/metawrap/config-metawrap_1.3.2 → docker/metawrap/config-metawrap
@@ -1,19 +1,14 @@
-#!/bin/bash
 # Paths to custon pipelines and scripts of metaWRAP
 mw_path=$(which metawrap)
 bin_path=${mw_path%/*}
 SOFT=${bin_path}/metawrap-scripts
 PIPES=${bin_path}/metawrap-modules
-
 # OPTIONAL databases (see 'Databases' section of metaWRAP README for details)
 # path to kraken standard databases
-KRAKEN_DB=~/KRAKEN_DB
-KRAKEN2_DB=~/KRAKEN_DB2
-
+KRAKEN_DB=/data2/KRAKEN_DB
+KRAKEN2_DB=/data2/KRAKEN_DB2
 # path to indexed human (or other host) genome (see metaWRAP website for guide). This includes .bitmask and .srprism files
-BMTAGGER_DB=~/BMTAGGER_DB
-
+BMTAGGER_DB=/data2/BMTAGGER_DB
 # paths to BLAST databases
-BLASTDB=~/NCBI_NT_DB
-TAXDUMP=~/NCBI_TAX_DB
-
+BLASTDB=/data2/NCBI_NT_DB
+TAXDUMP=/data2/NCBI_TAX_DB
diff --git a/docker/metawrap/docker_1.3.0 b/docker/metawrap/docker_1.3.0
diff --git a/docker/metawrap/docker_1.3.2 b/docker/metawrap/docker_1.3.2
diff --git a/docker/metawrap/mw b/docker/metawrap/mw
@@ -0,0 +1,5 @@
+#!/bin/bash
+set +eu
+source /etc/bash.bashrc
+/home/metaWRAP/bin/metawrap "$@"
+set -eu
diff --git a/docs/requirements.txt b/docs/requirements.txt
@@ -31,4 +31,5 @@ tornado==6.0.4
 tqdm==4.48.2
 zipp==3.1.0
 mkdocs-git-revision-date-plugin
+mkdocs-snakemake-rule-plugin
 mike