Merge pull request #63 from hoelzer/feat/species-sc2

Add SARS-CoV-2 as an easy to use species

Author: hoelzer

README.md

@@ -91,6 +91,7 @@ Currently supported are:
 |gga  | _Gallus gallus_      | [NCBI: Gallus_gallus.GRCg6a.dna.toplevel] |
 |cli  | _Columba livia_      | [NCBI: GCF_000337935.1_Cliv_1.0_genomic] |
 |eco  | _Escherichia coli_   | [Ensembl: Escherichia_coli_k_12.ASM80076v1.dna.toplevel] |
+|sc2  | _SARS-CoV-2_         | [ENA Sequence: MN908947.3 (Wuhan-Hu-1 complete genome) [web](https://www.ebi.ac.uk/ena/browser/view/MN908947.3) [fasta](https://www.ebi.ac.uk/ena/browser/api/fasta/MN908947.3?download=true)] |
 
 Included in this repository are:
 

clean.nf

@@ -79,7 +79,7 @@ if ( workflow.profile.contains('singularity') ) {
 }
 
 Set controls = ['phix', 'dcs', 'eno']
-Set hosts = ['hsa', 'mmu', 'cli', 'csa', 'gga', 'eco']
+Set hosts = ['hsa', 'mmu', 'cli', 'csa', 'gga', 'eco', 'sc2']
 Set input_types = ['nano', 'illumina', 'illumina_single_end', 'fasta']
 
 if ( params.profile ) { exit 1, "--profile is wrong, use -profile" }
@@ -231,7 +231,8 @@ def helpMSG() {
                                         - csa [NCBI: GCF_000409795.2_Chlorocebus_sabeus_1.1_genomic]
                                         - gga [NCBI: Gallus_gallus.GRCg6a.dna.toplevel]
                                         - cli [NCBI: GCF_000337935.1_Cliv_1.0_genomic]
-                                        - eco [Ensembl: Escherichia_coli_k_12.ASM80076v1.dna.toplevel]${c_reset}
+                                        - eco [Ensembl: Escherichia_coli_k_12.ASM80076v1.dna.toplevel]
+                                        - sc2 [ENA: MN908947.3 (Wuhan-Hu-1 complete genome)]${c_reset}
     ${c_green}--control${c_reset}       comma separated list of common controls used in Illumina or Nanopore sequencing [default: $params.control]
                                         ${c_dim}Currently supported are:
                                         - phix [Illumina: enterobacteria_phage_phix174_sensu_lato_uid14015, NC_001422]

modules/prepare_contamination.nf

@@ -40,6 +40,10 @@ process download_host {
     wget ftp://ftp.ensemblgenomes.org/pub/release-45/bacteria//fasta/bacteria_90_collection/escherichia_coli_k_12/dna/Escherichia_coli_k_12.ASM80076v1.dna.toplevel.fa.gz
     zcat *.gz | bgzip -@ ${task.cpus} -c > ${host}.fa.gz
   fi
+  if [ $host == 'sc2' ]; then
+    wget "https://www.ebi.ac.uk/ena/browser/api/fasta/MN908947.3?download=true" -O sc2.fa
+    cat sc2.fa | bgzip -@ ${task.cpus} -c > ${host}.fa.gz
+  fi
   """
   stub:
   """
@@ -108,4 +112,4 @@ process concat_contamination {
   """
   touch db.fa.gz db.fa.fai
   """
-}
+}

nextflow.config

@@ -104,6 +104,7 @@ profiles {
                 enabled = true
                 autoMounts = true
                 cacheDir = params.singularityCacheDir
+                envWhitelist = "HTTPS_PROXY,HTTP_PROXY,http_proxy,https_proxy,FTP_PROXY,ftp_proxy"
         }
         includeConfig 'configs/container.config'
     }