Merging the GenOMICC and UK Biobank Cohorts

This workflow yields combined:

• genotypes in PLINK bed format: obtained by subsampling the TOPMed UKB imputed genotypes to keep the GenOMICC variants.
• imputed genotypes in PLINK pgen format: ontained by merging UKB and GenOMICC TOPMed UKB imputed genotypes.
• covariates containing the current age and sex of individuals
• ancestry estimates: via the 1000 Genome Project

It is assumed that you have installed the required dependencies as explained in Working with the UKB RAP.

1. Uploading Inputs

This section describes the main inputs to the workflow that need to be uploaded to the RAP. I recommend deviating as little as possible from the following instructions to make sure things run smoothly afterwards. In this section we will populate the assets/rap folder in this repository that we will then upload to the RAP using the upload agent.

In particular we need to populate the genomicc and kgp subfolders as displayed below. The hesin_critical_fields.txt and main_fields.txt should already be present.

GenOMICC Data

We will need the following GenOMICC data, at the present time, the name of the file does not matter since it seems impossible to reference a file by its path on the RAP, instead we will have to use file IDs:

• genotypes: Output by Combining Datasets.
• covariates:
  • General covariates provided by Dominique, it should contain an age_years and a sex column.
  • Inferred covariates output by Combining Datasets containing ancestry estimates.
• imputed genotypes: Output by Genotypes Imputation.

I recommend to organise them as follows:

1000 Genome Project

You will also need the 1000 Genome Project genotypes in plink format, filtered to keep only variants matching the GenOMICC genotyped variants. This is an output of the Combining Datasets workflow. They can be organised as follows:

Other Assets

We also need the reference genome, which can be downloaded in the assets/rap folder with:

wget -O assets/rap/Homo_sapiens_assembly38.fasta https://storage.googleapis.com/genomics-public-data/resources/broad/hg38/v0/Homo_sapiens_assembly38.fasta

Uploading the data

Since this is a lot of data, we need to use the upload agent. Asumming ua is in your path, run:

export PROJECT_ID=PPP
export AUTH_TOKEN=XXX
ua --project $PROJECT_ID --auth-token $AUTH_TOKEN --folder /assets assets/rap/ --recursive

2. Extracting Phenotypes

At this point, it seems impossible (or very difficult) to extract phenotypes from a WDL workflow because the source dataset is not a regular file (see this or that). We thus have to first run a native DNA Nexus workflow to extract some preliminary data from the source dataset. First we build the workflow:

dx build rap_workflows/export_covariates

Then run it (replace the DATASET_RECORD_ID with the one corresponding to your project):

DATASET_RECORD_ID=record-J0pqJxjJZF8G55f99FF11JJ9
dx run -y \
-istage-J0vx360JpYQ0Jg1QJ5Zv0PFx.dataset_or_cohort_or_dashboard=$DATASET_RECORD_ID \
-istage-J0vx360JpYQ0Jg1QJ5Zv0PFx.field_names_file_txt=/assets/hesin_critical_fields.txt \
-istage-J0ygjB0JpYQJg4b985gqYkx6.dataset_or_cohort_or_dashboard=$DATASET_RECORD_ID \
-istage-J0ygjB0JpYQJg4b985gqYkx6.field_names_file_txt=/assets/main_fields.txt \
/export_covariates

You can monitor the workflow on the RAP, once finished you should have two outputs in the /export_covariates_outputs folder: $and$.

3. Merging Cohorts

First you need to compile the WDL workflow and upload it to the RAP, this can be done with the following:

export DX_COMPILER_PATH=/Users/olabayle/dxCompiler/dxCompiler-2.13.0.jar
export PROJECT_ID=project-J0pkqyQJpYQ133JG1p2J1qzv
java -jar $DX_COMPILER_PATH compile rap_workflows/ukb_merge/workflow.wdl -f -project $PROJECT_ID -folder /workflows -inputs rap_workflows/ukb_merge/inputs.json

where the DX_COMPILER_PATH and PROJECT_ID have to be set appropriately. The compiler might output some warnings like missing input for non-optional parameter but you can ignore these.

Then, you can run the workflow with the following command

dx run -y \
-f rap_workflows/ukb_merge/inputs.dx.json \
--destination /ukb_merge_outputs/ \
/workflows/merge_ukb_and_genomicc