Whole genome risk prediction of common disease in human pre-implantation embryos

This repository contains code for Kumar, et al. "Whole genome prediction of common disease in human pre-implantation embryos"

Preimplantation Genetic Testing (PGT) of in-vitro fertilized embryos has been proposed as a methodto reduce transmission of common disease; however, more comprehensive embryo geneticassessment, combining the effects of common variants and rare variants, remains unavailable. Here,we used a combination of molecular and statistical techniques to reliably infer inherited genomesequence in 110 embryos and model susceptibility across 12 common conditions. We observe agenotype accuracy of 99.0-99.4% at sites relevant to polygenic risk scoring in cases from Day-5embryo biopsies and 97.2-99.1% in cases from Day-3 embryo biopsies. Combining rare variants withPRS magnifies predicted differences across sibling embryos. For example, in a couple with apathogenic BRCA1 variant, we predicted a 15-fold difference in odds ratio across siblings whencombining versus a 4.5-fold or 3-fold difference with BRCA1 or PRS alone. Our findings may informthe discussion of utility and implementation of genome-based PGT in clinical practice.

Installation

Please install

• Python 3
• R 3.6.3

The following R packages

• ggpubr
• ggplot

Executing Code

Calculate coverage and accuracy of embryo prediction

cd ivf_retrospective_pub

bash ./scripts/get_prediction_accuracy.sh predicted_embryo.vcf \
							 triofile.vcf  \
							 sites_of_interest.vcf \
							 mother.vcf \
							 dad.vcf

Output files will be in

• accuracy.txt
• predicted_embryo.correct
• predicted_embryo.wrong
• predicted_embryo.other

accuracy.txt contains accuracy and coverage stastics genome-wide and for particular sites of interest. Other files contain the sites that are correct, wrong, or other (e.g. multiallelic or discarded).

Run the Example

Example data is provided for NA12878 chr1.

cd data/accuracy
gunzip *.gz # uncompress the vcf files. The script accepts .vcf, not .vcf.gz files
cd ../..  # go back to the repo directory
bash ./scripts/get_prediction_accuracy.sh <ivf_retrospective_pub path>/data/accuracy/predicted_embryo.chr1.vcf \
 					  <ivf_retrospective_pub path>/data/accuracy/triofile.chr1.vcf \
					  <ivf_retrospective_pub path>/data/accuracy/retro.prs.sorted.vcf \
					  <ivf_retrospective_pub path>/data/accuracy/mom.chr1.vcf \
					  <ivf_retrospective_pub>/data/accuracy/dad.chr1.vcf

where <ivf_retrospective_pub> is the path to this repository.

Generating Figures

Code for generating figures in the paper is provided in the figures directory

 Rscript figures/Fig1C.R
 Rscript figures/Fig2.R
 Rscript figures/SuppFig1.R
 Rscript figures/SuppFig2.R
 Rscript figures/SuppFigure3.R