Merge pull request #2 from warthmann/developer

Developer
mv-lab · Nov 18, 2019 · fc0e856 · fc0e856
2 parents e488147 + 06445ca
commit fc0e856
Show file tree

Hide file tree

Showing 11 changed files with 65 additions and 54 deletions.
diff --git a/.gitignore b/.gitignore
@@ -10,7 +10,9 @@ output/*
 !output/readme
 
 # rawdata (size problem)
-genomes_and_annotations/snpeffdata/
+genomes_and_annotations/*
+!genomes_and_annotations/readme
+
 
 # Notes and experiments
 todo

diff --git a/Snakefile b/Snakefile
@@ -19,4 +19,4 @@ rule all:
         rules.align.input,
         rules.varcall.input,
         rules.stats.input,
-        rules.snpeff.output,
+#        rules.snpeff.output,
diff --git a/config.yml b/config.yml
@@ -35,6 +35,7 @@ qc:
 mapping:
   aligners:
     - bwa
+    - ngm
   refs:
     - Sorghum_bicolor_NCBIv3.42
   samplesets:
@@ -65,14 +66,14 @@ LB-impute:
 denovodist:
   kwip_sketchsize: 1e9
   mash_sketchsize: 1e5
-  sourmash_sketchsize: 1e5
+  sourmash_sketchsize: 100000
   ksize: 21
-  kwip_sets: []
+  kwip_sets:
+    - all_samples
   mash_sets: # we can subset from everything for (sour)mash
     - all_samples
   sourmash_sets: []
 
-
 varcall:
   minmapq:
     bwa: 30
@@ -82,7 +83,7 @@ varcall:
   chunksize: 1000000
   aligners:
     - bwa
-    #- ngm
+    - ngm
   callers:
     #- mpileup
     - freebayes

diff --git a/envs/additional.yml b/envs/additional.yml
@@ -0,0 +1,12 @@
+#conda env update --name my_env --file additional.yml
+#name: paneuc
+channels:
+  - defaults
+  - bioconda
+  - conda-forge
+  - r
+dependencies:
+  - qualimap
+  - r-rgl
+  - r-devtools
+  - r-vcfr
diff --git a/envs/condaenv.yml b/envs/condaenv.yml
@@ -1,28 +1,29 @@
-name: paneuc
+#name: paneuc
 channels:
   - defaults
   - bioconda
   - conda-forge
-  - r
+#  - r
 dependencies:
   - python=3.*
   - snakemake
+  - yamlordereddictloader
   - bwa
   - freebayes
   - vt
   - samtools
   - bcftools
   - bedops
-  - qualimap
   - AdapterRemoval
   - seqhax
   - pigz
   - matplotlib =2.2
   - pandas =0.23
   - seaborn =0.8
-  - r-rgl
-  - r-devtools
   - rust
   - khmer
-  - r-ggplot2
-  - r-vcfr
+  - kwip
+  - mash
+  - pysam
+  - nextgenmap
+  - sourmash
diff --git a/genomes_and_annotations/readme b/genomes_and_annotations/readme
@@ -1,38 +1,6 @@
 genomes_and_annotations/
 ├── readme
-├── snpeffdata
-│   ├── genomes
-│   │   └── Sorghum_bicolor_NCBIv3.42.fa
-│   ├── Readme
-│   └── Sorghum_bicolor_NCBIv3.42
-│       ├── genes.gtf.gz
-│       ├── sequence.10.bin
-│       ├── sequence.1.bin
-│       ├── sequence.2.bin
-│       ├── sequence.3.bin
-│       ├── sequence.4.bin
-│       ├── sequence.5.bin
-│       ├── sequence.6.bin
-│       ├── sequence.7.bin
-│       ├── sequence.8.bin
-│       ├── sequence.9.bin
-│       ├── sequence.bin
-│       └── snpEffectPredictor.bin
 ├── Sorghum
-│   ├── Sorghum.fa
-│   ├── Sorghum.fa.amb
-│   ├── Sorghum.fa.ann
-│   ├── Sorghum.fa.bwt
-│   ├── Sorghum.fa.fai
-│   ├── Sorghum.fa.pac
-│   └── Sorghum.fa.sa
 └── Sorghum_bicolor_NCBIv3.42
-    ├── Sorghum_bicolor_NCBIv3.42.fa
-    ├── Sorghum_bicolor_NCBIv3.42.fa.amb
-    ├── Sorghum_bicolor_NCBIv3.42.fa.ann
-    ├── Sorghum_bicolor_NCBIv3.42.fa.bwt
-    ├── Sorghum_bicolor_NCBIv3.42.fa.fai
-    ├── Sorghum_bicolor_NCBIv3.42.fa.pac
-    └── Sorghum_bicolor_NCBIv3.42.fa.sa
 
-5 directories, 30 files
+2 directories, 1 file
diff --git a/rules/align.rules.smk b/rules/align.rules.smk
@@ -165,7 +165,7 @@ rule qualimap_samp:
     input:
         bam="output/alignments/samples/{aligner}/{ref}/{sample}.bam",
     output:
-        "output/alignments/qualimap/samples/{aligner}~{ref}~{sample}/",
+        directory("output/alignments/qualimap/samples/{aligner}~{ref}~{sample}/"),
     log:
         "output/log/align/qualimap_sample/{aligner}~{ref}~{sample}.log"
     threads: 4

diff --git a/rules/denovo.rules.smk b/rules/denovo.rules.smk
@@ -26,22 +26,31 @@ rule mash:
                sketchsize=config["denovodist"]["mash_sketchsize"],
                set=config["denovodist"]["mash_sets"]),
 
-rule pca:
+rule pca_mash:
     input:
-        rules.mash.input,
+        rules.mash.input, 
+    output:
+        "output/plots/denovo/mash/pca.pdf", 
+    script:
+        "../scripts/pca_mash.R"
+
+rule pca_kwip:
+    input:
+        rules.kwip.input,
     output:
-        "output/plots/denovo/mash/pca.pdf"
+        "output/plots/denovo/kwip/pca.pdf",
     script:
-        "../scripts/pca.R"
+        "../scripts/pca_kwip.R"
+
 
 
 rule denovo:
     input:
         rules.kwip.input,
         rules.mash.input,
         rules.sourmash.input,
-        rules.pca.output,
-
+        rules.pca_mash.output,
+	rules.pca_kwip.output,
 
 ##### Actual rules #####
 

diff --git a/rules/snpeff.rules.smk b/rules/snpeff.rules.smk
@@ -16,7 +16,7 @@ rule snpeff:
         "output/log/snpeff/snpeff.log"
     params:
         reference=config["snpeff"]['name'],
-        extra="-Xmx6g"
+        extra="-Xmx6g -v"
     wrapper:
         "0.27.1/bio/snpeff"
 

diff --git a/scripts/pca.R → scripts/pca_kwip.R b/scripts/pca.R → scripts/pca_kwip.R
@@ -5,8 +5,11 @@ print (snakemake@output[[1]])
 
 y0<-read.delim(snakemake@input[[1]], header=T)
 data<-as.matrix(y0[, -1])
+labels<-colnames(data)
 pca <- prcomp(data, scale=F)
 print ('PCA done!')
 plot3d(pca$x[,c(1,2,3)], size=10)
+text3d(pca$x[,c(1,2,3)], text=labels)
 rgl.postscript(snakemake@output[[1]],"pdf")
+rgl.postscript(snakemake@output[[1]],"svg")
 
diff --git a/scripts/pca_mash.R b/scripts/pca_mash.R
@@ -0,0 +1,15 @@
+library(rgl)
+
+print (snakemake@input[[1]])
+print (snakemake@output[[1]])
+
+y0<-read.delim(snakemake@input[[1]], header=T)
+data<-as.matrix(y0[, -1])
+labels<-as.data.frame(do.call(rbind, strsplit(colnames(data), "[.]")))$V4
+pca <- prcomp(data, scale=F)
+print ('PCA done!')
+plot3d(pca$x[,c(1,2,3)], size=10)
+text3d(pca$x[,c(1,2,3)], text=labels)
+rgl.postscript(snakemake@output[[1]],"pdf")
+rgl.postscript(snakemake@output[[1]],"svg")
+