Merge pull request #34 from MaxUlysse/AWS-iGenomes

update reference files
nf-core · Sep 13, 2019 · c0a86e2 · c0a86e2
2 parents 5ea919f + 4f6cc22
commit c0a86e2
Show file tree

Hide file tree

Showing 4 changed files with 92 additions and 55 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -15,7 +15,7 @@ Initial release of `nf-core/sarek`, created with the [nf-core](http://nf-co.re/)
 - [#2](https://github.com/nf-core/sarek/pull/2), [#3](https://github.com/nf-core/sarek/pull/3), [#4](https://github.com/nf-core/sarek/pull/4), [#5](https://github.com/nf-core/sarek/pull/5), [#7](https://github.com/nf-core/sarek/pull/7), [#9](https://github.com/nf-core/sarek/pull/9), [#10](https://github.com/nf-core/sarek/pull/10), [#11](https://github.com/nf-core/sarek/pull/11), [#12](https://github.com/nf-core/sarek/pull/12) - Add CI for `nf-core/sarek`
 - [#3](https://github.com/nf-core/sarek/pull/3) - Add preprocessing to `nf-core/sarek`
 - [#4](https://github.com/nf-core/sarek/pull/4) - Add variant calling to `nf-core/sarek` with `HaplotypeCaller`, and single mode `Manta` and `Strelka`
-- [#5](https://github.com/nf-core/sarek/pull/5) - Add variant calling to `nf-core/sarek` with `Manta`, `Strelka`, `Strelka Best Practices`, `MuTecT2`, `FreeBayes`, `ASCAT`, `ControlFREEC`
+- [#5](https://github.com/nf-core/sarek/pull/5), [#34](https://github.com/nf-core/sarek/pull/34) - Add variant calling to `nf-core/sarek` with `Manta`, `Strelka`, `Strelka Best Practices`, `MuTecT2`, `FreeBayes`, `ASCAT`, `ControlFREEC`
 - [#6](https://github.com/nf-core/sarek/pull/6) - Add default containers for annotation to `nf-core/sarek`
 - [#7](https://github.com/nf-core/sarek/pull/7) - Add MultiQC
 - [#7](https://github.com/nf-core/sarek/pull/7) - Add annotation
@@ -80,9 +80,10 @@ Initial release of `nf-core/sarek`, created with the [nf-core](http://nf-co.re/)
 - [#23](https://github.com/nf-core/sarek/pull/23) - `--genomeDict` is now deprecated, use `--dict` instead
 - [#24](https://github.com/nf-core/sarek/pull/24) - iGenomes config now contains germline resource for GATK4 Mutect2
 - [#30](https://github.com/nf-core/sarek/pull/30) - Simplify code for `MapReads` process
+- [#24](https://github.com/nf-core/sarek/pull/24) - iGenomes config now contains germline resource for `GATK4 Mutect2`
 - [#31](https://github.com/nf-core/sarek/pull/31) - Move extra CI to GitHub Actions nf-core extra CI
 - [#32](https://github.com/nf-core/sarek/pull/32), [#33](https://github.com/nf-core/sarek/pull/33) - Install `ASCAT` with `conda` in the `environment.yml` file
-- [#33](https://github.com/nf-core/sarek/pull/33) - use workflow.manifest.version to specify workflow version in path to R scripts for control-FREEC and VEP processes
+- [#33](https://github.com/nf-core/sarek/pull/33) - use `workflow.manifest.version` to specify workflow version in path to scripts for `ControlFREEC` and `VEP` processes
 
 ### `Removed`
 
@@ -111,6 +112,14 @@ Initial release of `nf-core/sarek`, created with the [nf-core](http://nf-co.re/)
 - [#31](https://github.com/nf-core/sarek/pull/31) - Fix rcolorbrewer version according to nf-core lint
 - [#33](https://github.com/nf-core/sarek/pull/33) - Fix MD Linting
 
+### `Deprecated`
+
+- [#23](https://github.com/nf-core/sarek/pull/23) - `--sample` is now deprecated, use `--input` instead
+- [#23](https://github.com/nf-core/sarek/pull/23) - `--genomeFile` is now deprecated, use `--fasta` instead
+- [#23](https://github.com/nf-core/sarek/pull/23) - `--genomeIndex` is now deprecated, use `--fastaFai` instead
+- [#23](https://github.com/nf-core/sarek/pull/23) - `--genomeDict` is now deprecated, use `--dict` instead
+- [#29](https://github.com/nf-core/sarek/pull/29) - `--noReports` is now deprecated, use `--skipQC all`
+
 ## [2.3.FIX1] - 2019-03-04
 
 ### `Fixed`

diff --git a/conf/genomes.config b/conf/genomes.config
@@ -10,31 +10,37 @@
 params {
   genomes {
     'GRCh37' {
-      acLoci           = "${params.genomes_base}/1000G_phase3_20130502_SNP_maf0.3.loci"
-      acLociGC         = "${params.genomes_base}/1000G_phase3_20130502_SNP_maf0.3.loci.gc"
-      bwaIndex         = "${params.genomes_base}/human_g1k_v37_decoy.fasta.{amb,ann,bwt,pac,sa}"
-      dbsnp            = "${params.genomes_base}/dbsnp_138.b37.vcf"
-      dbsnpIndex       = "${params.genomes_base}/dbsnp_138.b37.vcf.idx"
-      dict             = "${params.genomes_base}/human_g1k_v37_decoy.dict"
-      fasta            = "${params.genomes_base}/human_g1k_v37_decoy.fasta"
-      fastaFai         = "${params.genomes_base}/human_g1k_v37_decoy.fasta.fai"
-      intervals        = "${params.genomes_base}/wgs_calling_regions_Sarek.list"
-      knownIndels      = "${params.genomes_base}/{1000G_phase1,Mills_and_1000G_gold_standard}.indels.b37.vcf"
-      knownIndelsIndex = "${params.genomes_base}/{1000G_phase1,Mills_and_1000G_gold_standard}.indels.b37.vcf.idx"
-      snpeffDb         = "GRCh37.75"
-      vepCacheVersion  = "95"
+      acLoci                = "${params.genomes_base}/1000G_phase3_20130502_SNP_maf0.3.loci"
+      acLociGC              = "${params.genomes_base}/1000G_phase3_20130502_SNP_maf0.3.loci.gc"
+      bwaIndex              = "${params.genomes_base}/human_g1k_v37_decoy.fasta.{amb,ann,bwt,pac,sa}"
+      chrDir                = "${params.genomes_base}/Chromosomes"
+      chrLength             = "${params.genomes_base}/human_g1k_v37_decoy.len"
+      dbsnp                 = "${params.genomes_base}/dbsnp_138.b37.vcf"
+      dbsnpIndex            = "${params.genomes_base}/dbsnp_138.b37.vcf.idx"
+      dict                  = "${params.genomes_base}/human_g1k_v37_decoy.dict"
+      fasta                 = "${params.genomes_base}/human_g1k_v37_decoy.fasta"
+      fastaFai              = "${params.genomes_base}/human_g1k_v37_decoy.fasta.fai"
+      germlineResource      = "${params.genomes_base}/gnomAD.r2.1.1.GRCh37.PASS.AC.AF.only.vcf.gz"
+      germlineResourceIndex = "${params.genomes_base}/gnomAD.r2.1.1.GRCh37.PASS.AC.AF.only.vcf.gz.tbi"
+      intervals             = "${params.genomes_base}/wgs_calling_regions_Sarek.list"
+      knownIndels           = "${params.genomes_base}/{1000G_phase1,Mills_and_1000G_gold_standard}.indels.b37.vcf"
+      knownIndelsIndex      = "${params.genomes_base}/{1000G_phase1,Mills_and_1000G_gold_standard}.indels.b37.vcf.idx"
+      snpeffDb              = "GRCh37.75"
+      vepCacheVersion       = "95"
     }
     'GRCh38' {
       acLoci                = "${params.genomes_base}/1000G_phase3_GRCh38_maf0.3.loci"
       acLociGC              = "${params.genomes_base}/1000G_phase3_GRCh38_maf0.3.loci.gc"
       bwaIndex              = "${params.genomes_base}/Homo_sapiens_assembly38.fasta.64.{alt,amb,ann,bwt,pac,sa}"
+      chrDir                = "${params.genomes_base}/Chromosomes"
+      chrLength             = "${params.genomes_base}/Homo_sapiens_assembly38.len"
       dbsnp                 = "${params.genomes_base}/dbsnp_146.hg38.vcf.gz"
       dbsnpIndex            = "${params.genomes_base}/dbsnp_146.hg38.vcf.gz.tbi"
-      germlineResource      = "${params.genomes_base}/GCF_000001405.38.AUTOSOMESXY.COMMON.BIALLELIC.SNPs.with.AF.vcf.gz"
-      germlineResourceIndex = "${params.genomes_base}/GCF_000001405.38.AUTOSOMESXY.COMMON.BIALLELIC.SNPs.with.AF.vcf.gz.tbi"
       dict                  = "${params.genomes_base}/Homo_sapiens_assembly38.dict"
       fasta                 = "${params.genomes_base}/Homo_sapiens_assembly38.fasta"
       fastaFai              = "${params.genomes_base}/Homo_sapiens_assembly38.fasta.fai"
+      germlineResource      = "${params.genomes_base}/GCF_000001405.38.AUTOSOMESXY.COMMON.BIALLELIC.SNPs.with.AF.vcf.gz"
+      germlineResourceIndex = "${params.genomes_base}/GCF_000001405.38.AUTOSOMESXY.COMMON.BIALLELIC.SNPs.with.AF.vcf.gz.tbi"
       intervals             = "${params.genomes_base}/wgs_calling_regions.hg38.bed"
       knownIndels           = "${params.genomes_base}/{Mills_and_1000G_gold_standard.indels.hg38,beta/Homo_sapiens_assembly38.known_indels}.vcf.gz"
       knownIndelsIndex      = "${params.genomes_base}/{Mills_and_1000G_gold_standard.indels.hg38,beta/Homo_sapiens_assembly38.known_indels}.vcf.gz.tbi"

diff --git a/conf/igenomes.config b/conf/igenomes.config
@@ -7,41 +7,45 @@
  * path using $params.igenomes_base / --igenomes_base
  */
 
- params {
-   genomes {
-     'GRCh37' {
-       acLoci                 = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/ASCAT/1000G_phase3_20130502_SNP_maf0.3.loci"
-       acLociGC               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/ASCAT/1000G_phase3_20130502_SNP_maf0.3.loci.gc"
-       bwaIndex               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/BWAIndex/human_g1k_v37_decoy.fasta.{amb,ann,bwt,pac,sa}"
-       dbsnp                  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GATKBundle/dbsnp_138.b37.vcf"
-       dbsnpIndex             = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GATKBundle/dbsnp_138.b37.vcf.idx"
-       germlineResource       = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GermlineResource/gnomAD.r2.1.1.GRCh37.PASS.AC.AF.only.vcf.gz"
-       germlineResourceIndex  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GermlineResource/gnomAD.r2.1.1.GRCh37.PASS.AC.AF.only.vcf.gz.tbi"
-       dict                   = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/WholeGenomeFasta/human_g1k_v37_decoy.dict"
-       fasta                  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/WholeGenomeFasta/human_g1k_v37_decoy.fasta"
-       fastaFai               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/WholeGenomeFasta/human_g1k_v37_decoy.fasta.fai"
-       intervals              = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/intervals/wgs_calling_regions_Sarek.list"
-       knownIndels            = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GATKBundle/{1000G_phase1,Mills_and_1000G_gold_standard}.indels.b37.vcf"
-       knownIndelsIndex       = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GATKBundle/{1000G_phase1,Mills_and_1000G_gold_standard}.indels.b37.vcf.idx"
-       snpeffDb               = "GRCh37.75"
-       vepCacheVersion        = "95"
-     }
-     'GRCh38' {
-       acLoci                 = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/ASCAT/1000G_phase3_GRCh38_maf0.3.loci"
-       acLociGC               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/ASCAT/1000G_phase3_GRCh38_maf0.3.loci.gc"
-       bwaIndex               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/BWAIndex/Homo_sapiens_assembly38.fasta.64.{alt,amb,ann,bwt,pac,sa}"
-       dbsnp                  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/dbsnp_146.hg38.vcf.gz"
-       dbsnpIndex             = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/dbsnp_146.hg38.vcf.gz.tbi"
-       germlineResource       = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GermlineResource/gnomAD.r2.1.1.GRCh38.PASS.AC.AF.only.vcf.gz"
-       germlineResourceIndex  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GermlineResource/gnomAD.r2.1.1.GRCh38.PASS.AC.AF.only.vcf.gz.tbi"
-       dict                   = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/WholeGenomeFasta/Homo_sapiens_assembly38.dict"
-       fasta                  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/WholeGenomeFasta/Homo_sapiens_assembly38.fasta"
-       fastaFai               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/WholeGenomeFasta/Homo_sapiens_assembly38.fasta.fai"
-       intervals              = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/intervals/wgs_calling_regions.hg38.bed"
-       knownIndels            = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/{Mills_and_1000G_gold_standard.indels.hg38,beta/Homo_sapiens_assembly38.known_indels}.vcf.gz"
-       knownIndelsIndex       = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/{Mills_and_1000G_gold_standard.indels.hg38,beta/Homo_sapiens_assembly38.known_indels}.vcf.gz.tbi"
-       snpeffDb               = "GRCh38.86"
-       vepCacheVersion        = "95"
-     }
-   }
- }
+params {
+  genomes {
+    'GRCh37' {
+      acLoci                 = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/ASCAT/1000G_phase3_20130502_SNP_maf0.3.loci"
+      acLociGC               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/ASCAT/1000G_phase3_20130502_SNP_maf0.3.loci.gc"
+      bwaIndex               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/BWAIndex/human_g1k_v37_decoy.fasta.{amb,ann,bwt,pac,sa}"
+      chrDir                 = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/Chromosomes"
+      chrLength              = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/Length/human_g1k_v37_decoy.len"
+      dbsnp                  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GATKBundle/dbsnp_138.b37.vcf"
+      dbsnpIndex             = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GATKBundle/dbsnp_138.b37.vcf.idx"
+      dict                   = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/WholeGenomeFasta/human_g1k_v37_decoy.dict"
+      fasta                  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/WholeGenomeFasta/human_g1k_v37_decoy.fasta"
+      fastaFai               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Sequence/WholeGenomeFasta/human_g1k_v37_decoy.fasta.fai"
+      germlineResource       = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GermlineResource/gnomAD.r2.1.1.GRCh37.PASS.AC.AF.only.vcf.gz"
+      germlineResourceIndex  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GermlineResource/gnomAD.r2.1.1.GRCh37.PASS.AC.AF.only.vcf.gz.tbi"
+      intervals              = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/intervals/wgs_calling_regions_Sarek.list"
+      knownIndels            = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GATKBundle/{1000G_phase1,Mills_and_1000G_gold_standard}.indels.b37.vcf"
+      knownIndelsIndex       = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh37/Annotation/GATKBundle/{1000G_phase1,Mills_and_1000G_gold_standard}.indels.b37.vcf.idx"
+      snpeffDb               = "GRCh37.75"
+      vepCacheVersion        = "95"
+    }
+    'GRCh38' {
+      acLoci                 = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/ASCAT/1000G_phase3_GRCh38_maf0.3.loci"
+      acLociGC               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/ASCAT/1000G_phase3_GRCh38_maf0.3.loci.gc"
+      bwaIndex               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/BWAIndex/Homo_sapiens_assembly38.fasta.64.{alt,amb,ann,bwt,pac,sa}"
+      chrDir                 = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/Chromosomes"
+      chrLength              = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/Length/Homo_sapiens_assembly38.len"
+      dbsnp                  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/dbsnp_146.hg38.vcf.gz"
+      dbsnpIndex             = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/dbsnp_146.hg38.vcf.gz.tbi"
+      dict                   = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/WholeGenomeFasta/Homo_sapiens_assembly38.dict"
+      fasta                  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/WholeGenomeFasta/Homo_sapiens_assembly38.fasta"
+      fastaFai               = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Sequence/WholeGenomeFasta/Homo_sapiens_assembly38.fasta.fai"
+      germlineResource       = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GermlineResource/gnomAD.r2.1.1.GRCh38.PASS.AC.AF.only.vcf.gz"
+      germlineResourceIndex  = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GermlineResource/gnomAD.r2.1.1.GRCh38.PASS.AC.AF.only.vcf.gz.tbi"
+      intervals              = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/intervals/wgs_calling_regions.hg38.bed"
+      knownIndels            = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/{Mills_and_1000G_gold_standard.indels.hg38,beta/Homo_sapiens_assembly38.known_indels}.vcf.gz"
+      knownIndelsIndex       = "${params.igenomes_base}/Homo_sapiens/GATK/GRCh38/Annotation/GATKBundle/{Mills_and_1000G_gold_standard.indels.hg38,beta/Homo_sapiens_assembly38.known_indels}.vcf.gz.tbi"
+      snpeffDb               = "GRCh38.86"
+      vepCacheVersion        = "95"
+    }
+  }
+}
diff --git a/docs/usage.md b/docs/usage.md
@@ -28,6 +28,8 @@
   * [`--acLoci`](#--acloci)
   * [`--acLociGC`](#--aclocigc)
   * [`--bwaIndex`](#--bwaindex)
+  * [`--chrDir`](#--chrdir)
+  * [`--chrLength`](#--chrlength)
   * [`--dbsnp`](#--dbsnp)
   * [`--dbsnpIndex`](#--dbsnpindex)
   * [`--dict`](#--dict)
@@ -344,6 +346,22 @@ If you prefer, you can specify the full path to your reference genome when you r
 --bwaIndex '[path to the bwa indexes]'
 ```
 
+### `--chrDir`
+
+If you prefer, you can specify the full path to your reference genome when you run the pipeline:
+
+```bash
+--chrDir '[path to the Chromosomes folder]'
+```
+
+### `--chrLength`
+
+If you prefer, you can specify the full path to your reference genome when you run the pipeline:
+
+```bash
+--chrLength '[path to the Chromosomes length file]'
+```
+
 ### `--dbsnp`
 
 If you prefer, you can specify the full path to your reference genome when you run the pipeline: