Misc update prior to fishpond script for Salva

Author: guillaumecharbonnier

.gitignore

@@ -5,6 +5,7 @@ opt/
 log/
 bkp/
 shr/
+inp/
 inp/me/
 .snakemake/
 .nextflow/

src/chromhmm/test-srr.tsv

@@ -0,0 +1,8 @@
+name:
+  nanoplot
+channels:
+  - conda-forge
+  - bioconda
+  - nodefaults
+dependencies:
+  - nanoplot

src/snakemake/envs/nanoplot.yaml

@@ -0,0 +1,49 @@
+# A new environment for R kernel to share with the team
+name:
+  rkernel_main
+channels:
+  - conda-forge
+  - bioconda
+  - nodefaults
+dependencies:
+  - jupyter
+  - bioconductor-biomart
+  - bioconductor-chipseeker
+  - bioconductor-complexheatmap
+  - bioconductor-deseq2
+  - bioconductor-ensdb.hsapiens.v86
+  - bioconductor-ensembldb
+  - bioconductor-genomicranges
+  - bioconductor-org.hs.eg.db
+  - bioconductor-rtracklayer
+  - bioconductor-summarizedexperiment
+  - bioconductor-txdb.hsapiens.ucsc.hg19.knowngene
+  - bioconductor-txdbmaker
+  - bioconductor-fishpond
+  - bioconductor-tximeta
+  - bioconductor-rnbeads
+  - bioconductor-rnbeads.hg19
+  - bioconductor-rnbeads.hg38
+  - bioconductor-watermelon
+  - r-base
+  - r-arrow
+  - r-devtools
+  - r-biocmanager
+  - r-circlize
+  - r-data.table
+  - r-dplyr
+  - r-extrafont
+  - r-ggbeeswarm
+  - r-ggpubr
+  - r-ggrepel
+  - r-irkernel
+  - r-openxlsx
+  - r-rcolorbrewer
+  - r-readr
+  - r-readxl
+  - r-stringr
+  - r-tibble
+  - r-tidyr
+    #- r-tools # Not available in current channels
+  - r-vcfr
+  - r-xml2

src/snakemake/envs/rkernel_main.yaml

@@ -0,0 +1,13 @@
+name:
+  rnbeads
+channels:
+  - conda-forge
+  - bioconda
+  - nodefaults
+dependencies:
+  - r-ggplot2 <3.4 # https://github.com/epigen/RnBeads/issues/71
+  - bioconductor-rnbeads
+  - bioconductor-rnbeads.hg19
+  - bioconductor-rnbeads.hg38
+  - bioconductor-watermelon
+

src/snakemake/envs/rnbeads.yaml

@@ -29,6 +29,8 @@ rule agent_trim_pe:
         extra = params_extra
     wildcard_constraints:
         tool = "agent/trim"
+    resources:
+    	high_io = 1
     conda:
         "../envs/agent.yaml"
     shell:
@@ -73,6 +75,8 @@ rule agent_trim_2_lanes_pe:
         extra = params_extra
     wildcard_constraints:
         tool = "agent/trim_2_lanes"
+    resources:
+    	high_io = 1
     conda:
         "../envs/agent.yaml"
     shell:

src/snakemake/rules/agent_trimmer.smk

@@ -14,4 +14,5 @@ rule agent_wget:
         cd $OUTDIR
         wget 'https://dt4ei3l3hxs7z.cloudfront.net/?elqTrackId=30b3c5b8c3bd44f7b3a01b66ab2a30a5&elqaid=3928&elqat=2' --output-document 'AGeNT_2.0.5.zip'
         unzip AGeNT_2.0.5.zip
+	chmod +x {output.agent}
         """

src/snakemake/rules/agent_wget.smk

@@ -14,25 +14,25 @@ rule deepTools_computeMatrix_extra:
     """
     input:
         bw  = lambda wildcards: eval(mwconf['ids'][wildcards.bw_list_id]),
-        bed = lambda wildcards: eval(mwconf['ids'][wildcards.bed_list_id])
+        regions = lambda wildcards: eval(mwconf['ids'][wildcards.regions_list_id])
     output:
-        matrix = "out/{tool}{extra}_{bed_list_id}_{bw_list_id}.txt.gz"
+        matrix = "out/{tool}{extra}_{regions_list_id}_{bw_list_id}.txt.gz"
     log:
-                 "out/{tool}{extra}_{bed_list_id}_{bw_list_id}.log"
+                 "out/{tool}{extra}_{regions_list_id}_{bw_list_id}.log"
     benchmark:
-                 "out/{tool}{extra}_{bed_list_id}_{bw_list_id}.benchmark.tsv"
+                 "out/{tool}{extra}_{regions_list_id}_{bw_list_id}.benchmark.tsv"
     params:
         extra = params_extra
     conda:
         "../envs/deeptools.yaml"
     wildcard_constraints:
         tool="deepTools/computeMatrix",
         bw_list_id = "bw-[a-zA-Z0-9-]+",
-        bed_list_id = "bed-[a-zA-Z0-9-]+"
+        regions_list_id = "(bed|gtf)-[a-zA-Z0-9-]+"
     threads:
         MAX_THREADS
     shell:
-        "computeMatrix {params.extra} --regionsFileName {input.bed} "
+        "computeMatrix {params.extra} --regionsFileName {input.regions} "
         "--scoreFileName {input.bw} --outFileName {output.matrix} "
         "--numberOfProcessors {threads} &> {log}"
 

src/snakemake/rules/deepTools_computeMatrix.smk

@@ -14,7 +14,7 @@ rule ont_modkit_pileup_extra:
         bai="out/{filler}.bam.bai",
         fa= lambda wildcards: eval(mwconf['ids'][wildcards.fa_genome_id])
     output:
-        bed="out/{tool}{extra}_{fa_genome_id}/{filler}.bed"
+        bed="out/{tool}{extra}_{fa_genome_id}/{filler}.bed.gz"
     log:
             "out/{tool}{extra}_{fa_genome_id}/{filler}.log"
     benchmark:
@@ -29,5 +29,5 @@ rule ont_modkit_pileup_extra:
         MAX_THREADS
     shell:
         """
-        modkit pileup {params.extra} --threads {threads} --ref {input.fa} --log {log} {input.bam} {output.bed}
+        modkit pileup {params.extra} --threads {threads} --ref {input.fa} --log {log} {input.bam} - | bgzip > {output.bed}
         """

src/snakemake/rules/ont-modkit_pileup.smk

@@ -31,6 +31,9 @@ rule star_pe_extra:
         "../envs/star.yaml"
     threads:
         8
+    resources:
+        high_io = 1,
+        ram = 40
     shell:
         """
         ( WDIR=`pwd`
@@ -81,6 +84,9 @@ rule star_se_extra:
         "../envs/star.yaml"
     threads:
         8
+    resources:
+        high_io = 1,
+        ram = 40
     shell:
         """
         WDIR=`pwd`