Making virus base config from demo configs

- We did provide demo config files for several viruses
- Now they can be included with virus-base-config

config/README.md

@@ -52,6 +52,12 @@ Currently, the following _virus base config_ are available:
 - [hiv](hiv.yaml): provides HXB2 as a reference sequence for HIV, and sets the default aligner to _ngshmmalign_.
 - [sars-cov-2](sars-cov-2.yaml): provides NC\_045512.2 as a reference sequence for SARS-CoV-2, sets the default aligner to _bwa_ and sets the variant calling to be done against the reference instead of the cohort's consensus.
   In addition, a look-up for the recent versions of ARTIC protocol is provided; this makes it possible to set per-sample protocol in the sample table, and to turn on amplicon trimming (see [amplicon protocols](#amplicon-protocols)).
+- [rsvb](rsvb.yaml) config file for Human respiratory syncytial virus B (RSV-B), used to process Illumina RSV samples.
+- [h3n2_ha](h3n2_ha.yaml) config-file used for the analysis of H3N2 segment HA from wastewater Illumina data on SRA available through the SRA Run accession: [SRP385331](https://www.ebi.ac.uk/ena/browser/view/PRJNA856656)
+- [drosophila_c_virus](drosophila_c_virus.yaml) configuration used for the analysis of drosphila C virus (DCV) Illumina samples in Lezcano et al., Virus Evolution, 2023, doi:[10.1093/ve/vead074](https://doi.org/10.1093/ve/vead074), NCBI BioProject accession number [PRJNA993483](https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA993483)
+- [](herpes_simplex_virus_2.yaml) config-file used for the analysis of herpes simplex virus 2 (HSV-2) Illumina samples in Lezcano et al., Virus Evolution, 2023, doi:[10.1093/ve/vead074](https://doi.org/10.1093/ve/vead074). Analysed sample is from  López-Muñoz AD, Rastrojo A, Kropp KA, Viejo-Borbolla A, Alcamí A. "Combination of long- and short-read sequencing fully resolves complex repeats of herpes simplex virus 2 strain MS complete genome". _Microb Genom._ 2021 Jun;7(6). Sample accession number [ERR3278849](https://www.ebi.ac.uk/ena/browser/view/ERR3278849)
+- [polio](polio.yaml) config-file used for the analysis of poliovirus MinION samples. sample accession number: [ERR4027774]](https://www.ebi.ac.uk/ena/browser/view/ERR4027774) (Shaw et al., 2020, DOI: https://doi.org/10.1128/jcm.00920-20)
+- [mpxv](mpxv.yaml) Monkey pox virus
 
 ### configuration manual
 

config/drosophila_c_virus.yaml

@@ -1,3 +1,5 @@
+---
+name: Drosphila C virus
 # config-file used for the analysis of drosphila C virus Illumina samples
 # Lezcano et al., Virus Evolution, 2023, https://doi.org/10.1093/ve/vead074
 # NCBI BioProject accession number PRJNA993483
@@ -8,25 +10,14 @@ general:
 
 input:
     reference: resources/drosphila_c_virus/NC_001834.1.fasta
-    datadir: resources/samples/
     read_length: 100
-    samples_file: samples.tsv
     paired: true
 
 consensus_bcftools:
     max_coverage: 150000
 
+lofreq:
+    consensus: false
+
 snv:
   consensus: false
-  disk_mb: 1250
-  mem_mb: 35000
-  time_min: 6000
-  threads: 64
-
-output:
-    snv: true
-    local: true
-    global: false
-    visualization: false
-    diversity: false
-    QA: false

config/h3n2_ha.yaml

@@ -1,75 +1,23 @@
+---
+name: Influenza A virus subtype H3N2
 # config-file used for the analysis of H3N2 segment HA
 # config file was used to analysis wastewater Illumina data from SRA available
 # through the SRA Run accession: SRP385331
 
 general:
     aligner: bwa
     primers_trimmer: samtools
-    threads: 6
     snv_caller: lofreq
-    temp_prefix: ./temp
     preprocessor: skip
 
 input:
-    datadir: samples/
-    samples_file: samples.tsv
     reference: "{VPIPE_BASEDIR}/../resources/h3n2_ha/h3n2_ha.fasta"
     genes_gff: "{VPIPE_BASEDIR}/../resources/h3n2_ha/gffs/h3n2_ha.gff3"
     paired: true
     read_length: 151
 
-output:
-    datadir: results/
-    snv: True
-    local: True
-    global: False
-    visualization: False
-    QA: False
-    diversity: False
-
-gunzip:
-    mem: 100000
-
-extract:
-    mem: 100000
-
-preprocessing:
-    mem: 10000
-
-sam2bam:
-    mem: 5000
-
-ref_bwa_index:
-    mem: 65536
-
-bwa_align:
-    mem: 40690
-    threads: 8
-
-bowtie_align:
-    mem: 12288
-    threads: 6
-
-coverage:
-    mem: 131072
-    threads: 32
-    time: 60
-
-minor_variants:
-    mem: 16384
-    threads: 64
-
-coverage_intervals:
-    coverage: 0
-    mem: 2000
-    threads: 1
-
 lofreq:
     consensus: false
 
 snv:
     consensus: false
-    localscratch: $TMPDIR
-    time: 240
-    mem: 1024
-    threads: 64

config/herpes_simplex_virus_2.yaml

@@ -1,3 +1,5 @@
+---
+name: Herpes simplex Virus 2
 # config-file used for the analysis of herpes_simplex_virus_2 Illumina samples
 # Lezcano et al., Virus Evolution, 2023, https://doi.org/10.1093/ve/vead074
 # deletion analysis
@@ -12,23 +14,10 @@ input:
     reference: resources/herpes_simplex_virus_2/MK855052.1.fasta
     datadir: resources/samples/
     read_length: 250
-    samples_file: samples.tsv
     paired: true
 
 consensus_bcftools:
     max_coverage: 150000
 
 snv:
   consensus: false
-  disk_mb: 1250
-  mem_mb: 35000
-  time_min: 6000
-  threads: 64
-
-output:
-    snv: true
-    local: true
-    global: false
-    visualization: false
-    diversity: false
-    QA: false

config/polio.yaml

@@ -1,25 +1,15 @@
+---
+name: Poliovirus (MinION)
 # config-file used for the analysis of poliovirus MinION samples
 # sample accession number: ERR4027774 (Shaw et al., 2020, DOI: https://doi.org/10.1128/jcm.00920-20)
 
 general:
-    virus_base_config: ""
     aligner: minimap
     preprocessor: skip
 
 input:
     reference: resources/polio/AY560657.1.fasta
-    datadir: resources/samples/
-    samples_file: config/samples.tsv
     paired: false
 
-output:
-    trim_primers: false
-    snv: false
-    local: false
-    global: false
-    visualization: false
-    QA: false
-    diversity: false
-
 minimap_align:
     preset: map-ont

config/rsvb.yaml

@@ -1,27 +1,21 @@
+---
+name: Respiratory Syncytial Virus B
 # config file for Human respiratory syncytial virus B
 # config file is used to process Illumina RSV samples
 
 general:
-    virus_base_config: ""
     preprocessor: "prinseq"
     aligner: "bwa"
     primers_trimmer: "samtools"
 
 input:
-    datadir: "samples/"
-    samples_file: "samples.tsv"
-    read_length: 251
+    reference: "{VPIPE_BASEDIR}/../resources/rsvb/MT107528.1.fasta"
     primers_bedfile: "{VPIPE_BASEDIR}/../resources/rsvb/RSVB_primers_400_V2.1.bed"
     inserts_bedfile: "{VPIPE_BASEDIR}/../resources/rsvb/RSVB_inserts_400_V2.1.bed"
-    reference: "{VPIPE_BASEDIR}/../resources/rsvb/MT107528.1.fasta"
+    read_length: 251
 
-output:
-    datadir: "results"
-    trim_primers: true
-    snv: false
-    local: false
-    global: false
-snv:
-    consensus: false
 lofreq:
     consensus: false
+
+snv:
+    consensus: false