Nahrwhals Phasing

Phase a VCF into 1000 genomes SNPs by NYCG.

Table of Contents

1. Description
2. Workflow
3. Post-processing Steps
4. Usage

Description

The Snakefile takes as input a series of phased SNP VCFs (one per chromosome), and the NW output folders. It checks all aligned reads, assigns them a phase, and determines statistics for the most likely phase. It then carries out this re-phasing.

Final output, is a re-phased vcf in res/vcf.

Workflow

Given the different post-processing steps, running NWs in its entirety requires some steps:

(1) Running NW

1. Run the ntk-scan-snakemake pipeline which repeatedly calls Nahrwhals (found in another repo):

   ssh seneca
   cd /g/korbel/hoeps/projects/nahr/ntk_interesting_loci/ntk-scan-snakemake
   ## Edit configs 
   conda activate nahrwhals-big

   This will produce an all.tsv.

2. Convert all.tsv to all.vcf using:

   R
   nahrwhals::write_vcf('/path/to/all.tsv', '/path/to/all.vcf')

(2) Phasing with External Reference

If you also want to phase with respect to an external phasing, edit the config/Snake.config.json to indicate binaries for a lot of tools, and links to the reference fasta, NYGC VCF, and NAHRwhals output path and vcf. Files that have to be custom-made for this workflow are:

• region_file: (can sometimes be recycled from "intervals.txt" from Nahrwhals)

  chr1-103543499-103821114
  chr1-108148088-108537579
  chr1-109662781-109726958
  ...
  

• sample_file:

  GM19129
  GM19434
  HG00171
  ...
  

Ensure you have a working R environment loaded. If on Seneca, use:

```
conda deactivate # (a few times if necessary)
conda activate nahrwhals-big
```

Run the workflow with:

```
snakemake --cores n
```

Post-processing Steps

Merging BCF Files

After running the Snakefile, you need to post-process the results to merge the BCF files:

1. First, convert the NW_flipped_50000.vcf produced from the Snakemake part to a BCF file:

   bcftools view -O b NW_flipped_50000.vcf -o NW_flipped_50000.bcf
   tabix -p bcf NW_flipped_50000.bcf

2. Concatenate the individual allsample.vcfs produced from the Snakemake. (This is untested, maybe need some more flags or so)

   bcftools concat /path/to/nygc_subsets/*_allsamples* -O z -o allregions_allsamples.vcf.gz

3. Run the BCF merging script with the generated BCF files:

   ./merge_bcfs.sh custom_header.txt NW_flipped_50000.bcf allregions_allsamples.vcf.gz merged_output.vcf.gz