umr1283 / rain

Reference-based Ancestry INference

UMR 1283 - Project Template

You can install umr1283 with:

remotes::install_github("umr1283/umr1283")

The default project directory architecture from umr1283::create_project (compatible with RStudio Project Wizard) is as follow:

library(umr1283)
create_project(path = "my-project", analyst_name = "UMR1283")

tree -a my-project
#> my-project/
#> ├── data
#> ├── .devcontainer
#> │   ├── devcontainer.json
#> │   └── Dockerfile
#> ├── docs
#> ├── .gitignore
#> ├── logs
#> ├── my-project.Rproj
#> ├── outputs
#> ├── README.md
#> ├── renv
#> ├── reports
#> ├── .Rprofile
#> ├── scripts
#> │   ├── 00-targets.R
#> │   ├── _dependencies.R
#> │   └── tar-utils
#> ├── _targets
#> └── _targets.R
#> 
#> 10 directories, 9 files

---

Getting help

If you encounter a clear bug, please file a minimal reproducible example on GitHub.
For questions and other discussion, please open a discussion on GitHub.

rain's Issues

Remove vcftools dependency(only keep bcftools)

Add option for point size in ethnicity graph

subtitle error makes empty graph ethnicity_1kg.png

The ragg::agg_png step (making the "ethnicity_1kg.png") in compute_pca function fails due to an error in the subtitle command.
(application used R version 4.2.0 ; Rain version 0.6.0.9000)

The current parameter digits = 0 makes an error :

format(1234, big.mark = ",", digits = 0)

Error in prettyNum(.Internal(format(x, trim, digits, nsmall, width, 3L, :
invalid value 0 for 'digits' argument

But I suggest to fix it as format(1234, big.mark = ",", digits = 1) to get the expected text "1,234".

See line :

rain/R/compute_pca.R

Line 144 in 6db527c

format(nrow(data.table::fread(paste0(input_plink, ".bim"))), big.mark = ",", digits = 0),

Enhancement and refactoring

After using the function R/estimate_ethnicity.R (on exom sequencing data)
We have discussed about several enhancement to developed :

• #9
  (only keep bcftools)
• Add the option like "--missing-to-ref" for the step that merge the input VCFs. So this will make the possibility to assume genotypes at missing sites are 0/0. The merged vcf will hold much more variants filled as ref, instead of just variants that are all non ref in input population.
• Check whether it will be interesting to fill all the WT variants met in the reference panel (let's say 1kg) by completing the input vcf with all common WT variants ever seen in our sequencing data (by reading all cover files or the capture may be).
• Check LD clumping
• #7 (check alternative methods to draw a lot of points with ggpointdensity or scattermore)