BatchUPDetection

workflow to detect UPDs in large cohorts

Requirements

Unix packages:

• nextflow
• bcftools

python packages:

• pandas
• altair
• cyvcf2
• multiprocess
• openpyxl

Usage

1. Install all required packages and python libraries.
2. Collect all vcf-files to analyse in a specific folder (=vcf_folder)
3. Create a family file according to family_file_template.csv
4. Run nextflow pipeline as:

nextflow run scripts/batchUPdetection.nf \
--family_files path/to/family_file.csv \
--vcf_folder path/to/vcf_folder/ \
--outdir_roh path/to/roh_output/ \
--outdir_isec path/to/isec_output/ \
--outdir_results path/to/result_output/

Folders roh_output, isec_output and result_output will be used to write corresponding result files

Results

The workflow will create the following files:

• ROH files: output of bcftools roh
• isec files: output of bcftools isec
• a series of html files with interactive plots:
  • ...scatter.html: these plots show the ratio of maternal over paternal over maternal, maternal over non-maternal and paternal over non-paternal variants per chromosome vs ROH coverage. These plots can be used to identify UPDs.
• upd_finder_overview:

  • this sheet contains all chromosomes of all samples with:

    • file names (index, mother, father)
    • setup (single, duo, trio)
    • chromosome
    • perc_roh: fraction of the chromosome that is covered by ROHs
    • mat_over_pat: ratio of maternal over paternal variants
    • mat_over_notmat: ratio of maternal over non-maternal variants (for duos)
    • pat_over_mat: ratio of paternal over maternal variants
    • pat_over_notpat: ratio of paternal over non-paternal variants (for duos)
    • tags: all tags assigned to the respective chromosome
    • consanguinity: wether or not a consanguinity is suspected

Tagging and cutoffs

In order to identify UPDs and consanguinity, several tags with a variety of cutoffs are used:

Flag	Cut-Offs	Status/Interpretation
consanguinity unlikely	<3 chromosomes with >10% ROHs	-
consanguinity likely	>=3 chromosomes >10% ROHs	handle potential UPD flags with extra care
roh_high	> 70% covered by ROH	potential Isodisomy or deletions
roh_high_mixed	20-70% covered by ROH	potential Isodisomy or mixed UPD
inh_ratio_high	in duos >2; in trios > 5; inheritance ratio	potential heterodisomy
insufficient_snv	<200 SNVs/chr	number of SNVs is insufficient to reliably detect UPD features