refactor: switch modules to containers

CCBR · Mar 29, 2024 · 80ea1a3 · 80ea1a3
1 parent bd349c4
commit 80ea1a3
Show file tree

Hide file tree

Showing 6 changed files with 29 additions and 88 deletions.
diff --git a/workflow/rules/align.smk b/workflow/rules/align.smk
@@ -54,8 +54,7 @@ rule star1p:
         starindexdir=STAR_INDEX_DIR,
         alignTranscriptsPerReadNmax=TOOLS["star"]["alignTranscriptsPerReadNmax"],
         randomstr=str(uuid.uuid4()),
-    envmodules:
-        TOOLS["star"]["version"],
+    container: config['containers']["star"]
     threads: getthreads("star1p")
     shell:
         """
@@ -247,6 +246,7 @@ rule merge_SJ_tabs:
         filter2_noncanonical=config["star_1pass_filter_viruses_noncanonical"],
         filter2_unannotated=config["star_1pass_filter_viruses_unannotated"],
     threads: getthreads("merge_SJ_tabs")
+    container: config['containers']['base']
     shell:
         """
 set -exo pipefail
@@ -305,10 +305,7 @@ rule star2p:
         starindexdir=STAR_INDEX_DIR,
         alignTranscriptsPerReadNmax=TOOLS["star"]["alignTranscriptsPerReadNmax"],
         randomstr=str(uuid.uuid4()),
-    envmodules:
-        TOOLS["star"]["version"],
-        TOOLS["sambamba"]["version"],
-        TOOLS["samtools"]["version"],
+    container: config['containers']['star_ucsc_cufflinks']
     threads: getthreads("star2p")
     shell:
         """
@@ -480,10 +477,7 @@ rule star_circrnafinder:
         starindexdir=STAR_INDEX_DIR,
         alignTranscriptsPerReadNmax=TOOLS["star"]["alignTranscriptsPerReadNmax"],
         randomstr=str(uuid.uuid4()),
-    envmodules:
-        TOOLS["star"]["version"],
-        TOOLS["sambamba"]["version"],
-        TOOLS["samtools"]["version"],
+    container: config['containers']['star_ucsc_cufflinks']
     threads: getthreads("star_circrnafinder")
     shell:
         """
@@ -576,9 +570,7 @@ rule find_circ_align:
         peorse=get_peorse,
         find_circ_dir=FIND_CIRC_DIR,
         randomstr=str(uuid.uuid4()),
-    envmodules:
-        TOOLS["python27"]["version"],
-        TOOLS["samtools"]["version"], TOOLS['bowtie2']['version']
+    container: config['containers']["nclscan_py27"]
     threads: getthreads("find_circ_align")
     shell:
         """
@@ -659,9 +651,7 @@ rule estimate_duplication:
     params:
         sample="{sample}",
         memG=getmemG("estimate_duplication"),
-    envmodules:
-        TOOLS["picard"]["version"],
-        TOOLS["java"]["version"],
+    container: config['containers']["picard"]
     shell:
         """
 set -exo pipefail

diff --git a/workflow/rules/create_index.smk b/workflow/rules/create_index.smk
@@ -23,13 +23,7 @@ rule create_index:
         randomstr=str(uuid.uuid4()),
         nclscan_dir=config["nclscan_dir"],
         nclscan_config=config["nclscan_config"],
-    envmodules:
-        TOOLS["star"]["version"],
-        TOOLS["bwa"]["version"],
-        TOOLS["samtools"]["version"],
-        TOOLS["ucsc"]["version"],
-        TOOLS["cufflinks"]["version"],
-        TOOLS["python37"]["version"],
+    container: config['containers']['star_ucsc_cufflinks']
     threads: getthreads("create_index")
     shell:
         """
@@ -61,9 +55,6 @@ STAR \\
 bash {params.script2} {params.reffa} {params.refdir}/separate_fastas
 ls {params.refdir}/separate_fastas/*.fa | awk {AWK1} > {output.fastalst}
 # may have to create bowtie1 index here.. has to be a separate rule ... see below
-
-
-
 """
 
 
@@ -79,8 +70,7 @@ rule create_mapsplice_index:
         separate_fastas=join(REF_DIR, "separate_fastas"),
         ebwt=join(REF_DIR, "separate_fastas_index"),
     threads: getthreads("create_mapsplice_index")
-    container:
-        "docker://cgrlab/mapsplice2:latest"
+    container: "docker://cgrlab/mapsplice2:latest"
     shell:
         """
 set -exo pipefail
@@ -99,7 +89,7 @@ rule create_bwa_index:
         log=join(REF_DIR,"bwa_index.log")
     params:
         reffa=REF_FA
-    envmodules: TOOLS["bwa"]["version"]
+    container: config['containers']["base"]
     shell:"""
 set -exo pipefail
 refdir=$(dirname {params.reffa})
@@ -115,7 +105,7 @@ rule create_bowtie2_index:
         bt2=join(REF_DIR,"ref.1.bt2")
     params:
         reffa=REF_FA
-    envmodules: TOOLS["bowtie2"]["version"]
+    container: config['containers']["base"]
     shell:"""
 set -exo pipefail
 refdir=$(dirname {params.reffa})

diff --git a/workflow/rules/findcircrna.smk b/workflow/rules/findcircrna.smk
@@ -200,8 +200,7 @@ rule circExplorer:
         randomstr=str(uuid.uuid4()),
         # script=join(SCRIPTS_DIR, "circExplorer_get_annotated_counts_per_sample.py"),  # this produces an annotated counts table to which counts found in BAMs need to be appended
     threads: getthreads("circExplorer")
-    envmodules:
-        TOOLS["circexplorer"]["version"],
+    container: config['containers']["circexplorer"]
     shell:
         """
 set -exo pipefail
@@ -285,9 +284,7 @@ rule ciri:
         script=join(SCRIPTS_DIR, "filter_ciriout.py"),
         randomstr=str(uuid.uuid4()),
     threads: getthreads("ciri")
-    envmodules:
-        TOOLS["bwa"]["version"],
-        TOOLS["samtools"]["version"],
+    container: config['containers']['base']
     shell:
         """
 set -exo pipefail
@@ -377,8 +374,7 @@ rule circExplorer_bwa:
         randomstr=str(uuid.uuid4()),
         # script=join(SCRIPTS_DIR, "circExplorer_get_annotated_counts_per_sample.py"),  # this produces an annotated counts table to which counts found in BAMs need to be appended
     threads: getthreads("circExplorer")
-    envmodules:
-        TOOLS["circexplorer"]["version"],
+    container: config['containers']["circexplorer"]
     shell:
         """
 set -exo pipefail
@@ -427,8 +423,7 @@ rule create_ciri_count_matrix:
         lookup=ANNOTATION_LOOKUP,
         outdir=join(WORKDIR, "results"),
         hostID=HOST + "ID",
-    envmodules:
-        TOOLS["python37"]["version"],
+    container: config['containers']['base']
     shell:
         """
 set -exo pipefail
@@ -459,8 +454,7 @@ rule create_circexplorer_count_matrix:
         lookup=ANNOTATION_LOOKUP,
         outdir=join(WORKDIR, "results"),
         hostID=HOST + "ID",
-    envmodules:
-        TOOLS["python37"]["version"],
+    container: config['containers']['base']
     shell:
         """
 cd {params.outdir}
@@ -510,8 +504,7 @@ rule clear:
         quantfile=join(WORKDIR, "results", "{sample}", "CLEAR", "quant.txt"),
     params:
         genepred=rules.create_index.output.genepred_w_geneid,
-    container:
-        "docker://nciccbr/ccbr_clear:latest"
+    container: config['containers']['clear']
     threads: getthreads("clear")
     shell:
         """
@@ -690,8 +683,7 @@ rule dcc:
             "{sample}.dcc.counts_table.tsv.filtered",
         ),
     threads: getthreads("dcc")
-    envmodules:
-        TOOLS["python27"]["version"],
+    container: config['containers']['dcc']
     params:
         peorse=get_peorse,
         dcc_strandedness=config["dcc_strandedness"],
@@ -857,8 +849,7 @@ rule mapsplice:
         outdir=join(WORKDIR, "results", "{sample}", "MapSplice"),
         randomstr=str(uuid.uuid4()),
     threads: getthreads("mapsplice")
-    container:
-        "docker://cgrlab/mapsplice2:latest"
+    container: config['containers']['mapsplice']
     shell:
         """
 set -exo pipefail
@@ -951,9 +942,7 @@ rule mapsplice_postprocess:
         bai=join(
             WORKDIR, "results", "{sample}", "MapSplice", "{sample}.mapsplice.cram.crai"
         ),
-    envmodules:
-        TOOLS["samtools"]["version"],
-        TOOLS["python27"]["version"],
+    container: config['containers']["nclscan_py27"]
     params:
         script=join(SCRIPTS_DIR, "create_mapsplice_per_sample_counts_table.py"),
         memG=getmemG("mapsplice_postprocess"),
@@ -1043,8 +1032,7 @@ rule nclscan:
             "NCLscan",
             "{sample}.nclscan.counts_table.tsv.filtered",
         ),
-    envmodules:
-        TOOLS["ncl_required_modules"],
+    container: config['containers']["nclscan_py27"]
     threads: getthreads("nclscan")
     params:
         workdir=WORKDIR,
@@ -1150,9 +1138,7 @@ rule circrnafinder:
         ),
         bsj_min_nreads=config["minreadcount"],
         randomstr=str(uuid.uuid4()),
-    envmodules:
-        TOOLS["perl"]["version"],
-        TOOLS["samtools"]["version"],
+    container: config['containers']['base']
     shell:
         """
 set -exo pipefail
@@ -1228,11 +1214,7 @@ rule find_circ:
         min_reads=config['circexplorer_bsj_circRNA_min_reads'],
         collapse_script=join(SCRIPTS_DIR,"_collapse_find_circ.py"),
         randomstr=str(uuid.uuid4()),
-    envmodules:
-        TOOLS["python27"]["version"],
-        TOOLS["bowtie2"]["version"],
-        TOOLS["samtools"]["version"],
-        TOOLS["parallel"]["version"],
+    container: config['containers']["nclscan_py27"]
     threads: getthreads("find_circ")
     shell:
         """
@@ -1576,8 +1558,7 @@ rule create_master_counts_file:
         resultsdir=join(WORKDIR, "results"),
         lookup_table=ANNOTATION_LOOKUP,
         bsj_min_nreads=config["circexplorer_bsj_circRNA_min_reads"],
-    envmodules:
-        TOOLS["python37"]["version"],
+    container: config['containers']['base']
     shell:
         """
 set -exo pipefail

diff --git a/workflow/rules/init.smk b/workflow/rules/init.smk
@@ -233,10 +233,6 @@ for sample in SAMPLES:
 # print(SAMPLESDF)
 # sys.exit()
 
-## Load tools from YAML file
-with open(config["tools"]) as f:
-    TOOLS = yaml.safe_load(f)
-
 ## Load cluster.json
 with open(config["cluster"]) as json_file:
     CLUSTER = yaml.safe_load(json_file)

diff --git a/workflow/rules/post_align_processing.smk b/workflow/rules/post_align_processing.smk
@@ -111,8 +111,7 @@ rule merge_genecounts:
     params:
         outdir=join(WORKDIR, "results"),
         rscript=join(SCRIPTS_DIR, "merge_ReadsPerGene_counts.R"),
-    envmodules:
-        TOOLS["R"]["version"],
+    container: config['containers']["R"]
     shell:
         """
 set -exo pipefail

diff --git a/workflow/rules/post_findcircrna_processing.smk b/workflow/rules/post_findcircrna_processing.smk
@@ -62,11 +62,7 @@ rule create_circExplorer_BSJ_bam:
         flankscript=join(SCRIPTS_DIR, "_append_splice_site_flanks_to_BSJs.py"),
         bam2bwscript=join(SCRIPTS_DIR, "bam_to_bigwig.sh"),
         randomstr=str(uuid.uuid4()),
-    envmodules:
-        TOOLS["python37"]["version"],
-        TOOLS["samtools"]["version"],
-        TOOLS["bedtools"]["version"],
-        TOOLS["ucsc"]["version"],
+    container: config['containers']["ucsc"]
     threads: getthreads("create_circExplorer_BSJ_bam")
     shell:
         """
@@ -218,12 +214,7 @@ rule create_circExplorer_linear_spliced_bams:
         # bam2bwscript=join(SCRIPTS_DIR,"bam_to_bigwig.sh"),
         outdir=join(WORKDIR, "results", "{sample}", "circExplorer"),
         randomstr=str(uuid.uuid4()),
-    envmodules:
-        TOOLS["python37"]["version"],
-        TOOLS["samtools"]["version"],
-        TOOLS["bedtools"]["version"],
-        TOOLS["ucsc"]["version"],
-        TOOLS["parallel"]["version"],
+    container: config['containers']["base"]
     threads: getthreads("create_circExplorer_linear_spliced_bams")
     shell:
         """
@@ -371,9 +362,7 @@ if RUN_MAPSPLICE:
             bash2nreads_pyscript=join(SCRIPTS_DIR, "_bam_get_alignment_stats.py"),
             randomstr=str(uuid.uuid4()),
         threads: getthreads("alignment_stats")
-        envmodules:
-            TOOLS["python37"]["version"],
-            TOOLS["parallel"]["version"],
+        container: config['containers']["base"]
         shell:
             """
     set -exo pipefail
@@ -428,9 +417,7 @@ else:
             bash2nreads_pyscript=join(SCRIPTS_DIR, "_bam_get_alignment_stats.py"),
             randomstr=str(uuid.uuid4()),
         threads: getthreads("alignment_stats")
-        envmodules:
-            TOOLS["python37"]["version"],
-            TOOLS["parallel"]["version"],
+        container: config['containers']["base"]
         shell:
             """
     set -exo pipefail
@@ -520,9 +507,7 @@ rule create_hq_bams:
         host=HOST,
         additives=ADDITIVES,
         viruses=VIRUSES,
-    envmodules:
-        TOOLS["python37"]["version"],
-        TOOLS["samtools"]["version"],
+    container: config['containers']["base"]
     shell:"""
 set -exo pipefail
 outdir=$(dirname {output.outbam})