AS-IBDNe

Snakemake pipeline for running ancestry-specific IBDNe

1. Set Up:

Creating conda environment (first time only):

source /share/hennlab/progs/miniconda3/etc/profile.d/conda.sh
conda create -n IBDne-env
conda activate IBDne-env
conda install -c bioconda htslib tabix bcftools rfmix
conda install pandas
conda install -c r r-magrittr r-doparallel
conda install -c conda-forge matplotlib brewer2mpl
conda install -c dranew shapeit

Activating conda environment (before running pipeline)

source /share/hennlab/progs/miniconda3/etc/profile.d/conda.sh
conda activate IBDne-env

2. Required input files and structure of pipeline directory

Please create a working directory to run the Snakefile in. The structure of the directory should be as follows. Items marked by asterisk are required by the snakefile to be located where they are specified. The location of non-asterisked items are passed to the Snakefile by the user, so their location in the working directory is flexible.

|__ Snakefile*
|__ config.yaml
|__ data*
    |__ {dataset}.bim*
    |__ {dataset}.bed*
    |__ {dataset}.fam*
    |__ ref_smpmap.txt
    |__ admixed_ids.txt
    |__ centromeres_hg19.bed*
|__ scripts*
    |__ shapeit_to_germline.py*
    |__ msp_to_vit.py*
    |__ filter_gapfilled_ibd_ancestry.py*
    |__ reformat_vit.sh*
    |__ sep_anc_comb_chr.sh*
    |__ adjust_npairs.py*
    |__ plot_karyogram.R*
    |__ plot_ibdne.R*
    |__ shapeit_iterate.sh*
    |__ trim_bimfile.R*
|__ progs*
    |__ ibdne.05May18.1c3.jar*
    |__ filtercolumns.jar*
    |__ refined-ibd.17Jan20.102.jar*

The input dataset to this pipeline is a single bim/bed/fam fileset containing both the admixed individuals and the reference individuals.

3. QC beforehand

plink --bfile nama_tgp --geno 0.05 --mind 0.1 --make-bed --out nama_tgp_qc
# with populations in family ID column --> nama_tgp_qc_pops

4. Config file

The config file will contain all the file paths that change with every run of the snakemake pipeline.

Example file:

dataset: americans_subset_remove

rfmix_genmap: /share/hennlab/reference/recombination_maps/rfmix_combined_b37.map

smpmap: data/ref_smpmap.txt

admix_samples: data/admixed_ids.txt

ref: /share/hennlab/reference/1000G_Phase3_haps-sample-legend/1000GP_Phase3/1000GP_Phase3

chr_gmap: /share/hennlab/reference/recombination_maps/genetic_map_HapMapII_GRCh37/

mincM: 2

colors: green,purple,red,blue

Explanation of config input parameters:

• dataset: Name of the dataset. needs to match prefix of input bim/bed/fam files : {dataset}.bed /.bim /.fam
• rfmix_genmap: genetic map for all chromosomes. 3 tab delimited columns. 1- chromsome, 2- bp position, 3-recombination rate
• smpmap: contains all reference individuals. tab delimited file with IDs in first column and ancestry in second.
• admix_samples: list of admixed sample IDs. 1 column
• ref: prefix not including chromosome number and path to 1000 genome reference .hap.gz, .legend.gz and .sample files. for example, provide this:

  ref: /share/hennlab/reference/1000G_Phase3_haps-sample-legend/1000GP_Phase3/1000GP_Phase3

  if the files are named as:

  /share/hennlab/reference/1000G_Phase3_haps-sample-legend/1000GP_Phase3/1000GP_Phase3_chr1.legend.gz

• chr_gmap: path to genetic map files for individual chromosomes. 3 space delimited columns. 1-position bp, 2-Rate(cM/Mb) 3- Map(cM)
• mincM: minimum cM length of IBD segments to use in IBDne program
• colors: comma delimited list of colors for the output plots (karyograms and ibdne) to take on, in alphabetical order of ancestry. For example, if the reference populations are CHB,GBR,LWK,NAMA then the colors “green,purple,red,blue” will be assigned as CHB=green, GBR=purple, LWK=red, NAMA=blue.

5. Run Snakemake

# Dry run: always run first with -n flag to make sure the workflow will execute properly
nice /share/hennlab/progs/miniconda3/bin/snakemake --configfile config.yaml -j 20 -n
# Generate DAG
/share/hennlab/progs/miniconda3/bin/snakemake --configfile config.yaml -j 20 -n --rulegraph | dot -Tpng > rulegraph.png
# Run pipeline
nice /share/hennlab/progs/miniconda3/bin/snakemake --configfile config.yaml -j 20

6. Checking concordance between rfmix and admixture

If your reference panel has more than 3 populations, its recommended that you run admixture and compare it to the rfmix results as a check. To do this, use Alicia Martin’s script lai_global.py to convert the bedfiles output by the rule msp_to_bed and compare to the admixture output Q files.

Alicia’s script: https://github.com/armartin/ancestry_pipeline#estimate-global-ancestry-proportions-from-local-ancestry-inference

Admixture tutorial: https://github.com/hennlab/training/blob/main/common_analyses/admixture_pong.md

7. Output plots

This pipeline will output rfmix karyogram plots for every admixed individual, and 1 ibdne plot showing effective population size estimates for each reference population

8. Acknowledgements and sources:

• Rfmix version 2.3: https://github.com/slowkoni/rfmix/blob/master/MANUAL.md
• Scripts for processing rfmix 1.5 output to IBDne input: https://faculty.washington.edu/sguy/asibdne/
• IBDne: http://faculty.washington.edu/browning/ibdne.html
• filtercolumns.jar https://faculty.washington.edu/browning/beagle_utilities/utilities.html
• Script to convert msp format to bed format: Gerald Van Eeden