Merge pull request #4 from AnimalGenomicsETH/streamline

simplify config and expand readme

.test/config/config.yaml

@@ -24,31 +24,20 @@ frac_missing: .15
 ## Pangenome genotyping
 small_variants: DV.vcf.gz
 
-## Pangenie stuff
-#panel: '/cluster/work/pausch/alex/eQTL_GWAS/pangenie_panel.vcfwave.vcf'
 fastq: ''
 
-HiFi_samples:
-  sample1: hifi_1.fq.gz
-  sample2: hifi_2.fq.gz
-
-#pangenie_out: '/cluster/work/pausch/alex/eQTL_GWAS/pangenie_wave/samples.all.pangenie_genotyping.vcf.gz'
+samples:
+ - sample1
+ - sample2
 
-## Phasing and imputation
-#beagle4: ''
-beagle5: ''
-window: 10000000
-overlap: 20000
 
+## Variant comparison
 
-samples:
-  - sample1
-  - sample2
+HiFi_samples:
+  sample1: hifi_1.fq.gz
+  sample2: hifi_2.fq.gz
 
-## molecular QTL mapping
-#variants:
-#  PanGenie: '/cluster/work/pausch/alex/eQTL_GWAS/pangenie_8_wave/samples.all.pangenie_genotyping_DV.vcf.gz'
-#  SR: '/cluster/work/pausch/alex/eQTL_GWAS/variants/DV-SR/cohort.autosomes.WGS.imputed.vcf.gz'
+## Association testing
 
 covariates:
   eQTL:
@@ -63,5 +52,5 @@ mol_QTLs:
 
 
 permutations: 2500
-window: 1000000 #1 Mb
+window: 1000000 #1 Mb cis window
 chunks: 40

README.md

@@ -13,15 +13,23 @@ There are broadly three phases
   - Variant analysis (statistics, linkage disequibrium, SV overlap, etc.) 
   - Association mapping of e/sQTL
 
+
+An example of the input needed is given in the `config/example.yaml`, broadly requiring
+  - haplotype-resolved assemblies for pangenome panel creation
+  - small variants to supplement pangenome panel
+  - any HiFi samples to test SV completeness
+  - gene expression/splicing files and covariates for molecular QTL mapping
+
 Running with
 ```
-snakemake --configfile config/config.yaml
+snakemake --configfile config/example.yaml
 ```
-Will produce execute the following DAG
+Will execute the following DAG
 
 ![workflow](https://github.com/AnimalGenomicsETH/pangenome_molQTL/assets/29678761/bb0c73ca-fc31-4319-95e2-485da93f655a)
 
-which produces the major output files (e.g., accuracy comparison of PanGenie vs DeepVariant, SV overlap with Jasmine, conditional QTL analysis with QTLtools, etc.), which can then be independently analysed further.
+producing the major output files (e.g., accuracy comparison of PanGenie vs DeepVariant, SV overlap with Jasmine, conditional QTL analysis with QTLtools, etc.), which can then be independently analysed further.
+Many of these steps are computationally intensive, especially with many samples to genotype, and so effectively require some form of HPC.
 
 ### Citation
 

config/example.yaml

@@ -0,0 +1,58 @@
+## Pangenome panel
+reference: '/path/to/reference.fasta'
+
+assemblies:
+  asm1:
+    - 'asm1.hap1.fa
+    - 'asm1.hap2.fa'
+  asm2:
+    - 'asm2.hap1.fa'
+    - 'asm2.hap1.fa'
+  asm3:
+    - 'asm3.hap1.fa'
+    - 'asm3.hap2.fa'
+  asm4:
+    - 'asm4.hap1.fa'
+    - 'asm4.hap2.fa'
+trios:
+  asm1:
+    - asm2
+    - asm3
+
+scripts: '/path/to/PanGenie/scripts'
+outdir: 'pangenie_panel/'
+frac_missing: .15
+
+## Pangenome genotyping
+small_variants: DV.vcf.gz
+
+fastq: ''
+
+samples:
+ - sample1
+ - sample2
+
+## Variant comparison
+
+HiFi_samples:
+  sample1: hifi_1.fq.gz
+  sample2: hifi_2.fq.gz
+
+
+## molecular QTL mapping
+
+covariates:
+  eQTL:
+    Testis: 'eQTL.covar'
+  sQTL:
+    Testis: 'sQTL.covar'
+mol_QTLs:
+  eQTL:
+    Testis: 'eQTL.TPM'
+  sQTL:
+    Testis: 'sQTL.clusters'
+
+
+permutations: 2500
+window: 1000000 #1 Mb cis window
+chunks: 40