GCP Variant Transforms

Overview

This is a tool for transforming and processing VCF files in a scalable manner based on Apache Beam using Dataflow on Google Cloud Platform.

It can be used to directly load VCF files to BigQuery supporting hundreds of thousands of files, millions of samples, and billions of records. Additionally, it provides a preprocess functionality to validate the VCF files such that the inconsistencies can be easily identified.

Please see this presentation for a high level overview of BigQuery and how to effectively use Variant Transforms and BigQuery. Please also read the blog post about how a GCP customer used Variant Transforms for breakthrough clinical data science with BigQuery.

Prerequisites

1. Follow the getting started instructions on the Google Cloud page.
2. Enable the Genomics, Compute Engine, Cloud Storage, and Dataflow APIs
3. Create a new BigQuery dataset by visiting the BigQuery web UI, clicking on the down arrow icon next to your project name in the navigation, and clicking on Create new dataset.

Loading VCF files to BigQuery

Using docker

The easiest way to run the VCF to BigQuery pipeline is to use the docker image and run it with the Google Genomics Pipelines API as it has the binaries and all dependencies pre-installed. Please ensure you have the latest gcloud tool by running gcloud components update (more details here).

Run the script below and replace the following parameters:

• GOOGLE_CLOUD_PROJECT: This is your project ID that contains the BigQuery dataset.
• INPUT_PATTERN: A location in Google Cloud Storage where the VCF file are stored. You may specify a single file or provide a pattern to load multiple files at once. Please refer to the Variant Merging documentation if you want to merge samples across files. The pipeline supports gzip, bzip, and uncompressed VCF formats. However, it runs slower for compressed files as they cannot be sharded.
• OUTPUT_TABLE: The full path to a BigQuery table to store the output.
• TEMP_LOCATION: This can be any folder in Google Cloud Storage that your project has write access to. It's used to store temporary files and logs from the pipeline.

#!/bin/bash
# Parameters to replace:
GOOGLE_CLOUD_PROJECT=GOOGLE_CLOUD_PROJECT
INPUT_PATTERN=gs://BUCKET/*.vcf
OUTPUT_TABLE=GOOGLE_CLOUD_PROJECT:BIGQUERY_DATASET.BIGQUERY_TABLE
TEMP_LOCATION=gs://BUCKET/temp

COMMAND="/opt/gcp_variant_transforms/bin/vcf_to_bq \
  --project ${GOOGLE_CLOUD_PROJECT} \
  --input_pattern ${INPUT_PATTERN} \
  --output_table ${OUTPUT_TABLE} \
  --temp_location ${TEMP_LOCATION} \
  --job_name vcf-to-bigquery \
  --runner DataflowRunner"
gcloud alpha genomics pipelines run \
  --project "${GOOGLE_CLOUD_PROJECT}" \
  --logging "${TEMP_LOCATION}/runner_logs_$(date +%Y%m%d_%H%M%S).log" \
  --zones us-west1-b \
  --service-account-scopes https://www.googleapis.com/auth/cloud-platform \
  --docker-image gcr.io/gcp-variant-transforms/gcp-variant-transforms \
  --command-line "${COMMAND}"

Please note the operation ID returned by the above script. You can track the status of your operation by running:

gcloud alpha genomics operations describe <operation-id>

The returned data will have done: true when the operation is done. A detailed description of the Operation resource can be found in the API documentation.

The underlying pipeline uses Cloud Dataflow. You can navigate to the Dataflow Console, to see more detailed view of the pipeline (e.g. number of records being processed, number of workers, more detailed error logs).

Running from github

In addition to using the docker image, you may run the pipeline directly from source. First install git, python, pip, and virtualenv:

sudo apt-get install -y git python-pip python-dev build-essential
sudo pip install --upgrade pip
sudo pip install --upgrade virtualenv

Run virtualenv, clone the repo, and install pip packages:

virtualenv venv
source venv/bin/activate
git clone https://github.com/googlegenomics/gcp-variant-transforms.git
cd gcp-variant-transforms
pip install --upgrade .

You may use the DirectRunner (aka local runner) for small (e.g. 10,000 records) files or DataflowRunner for larger files. Files should be stored on Google Cloud Storage if using Dataflow, but may be stored locally for DirectRunner.

Example command for DirectRunner:

python -m gcp_variant_transforms.vcf_to_bq \
  --input_pattern gcp_variant_transforms/testing/data/vcf/valid-4.0.vcf \
  --output_table GOOGLE_CLOUD_PROJECT:BIGQUERY_DATASET.BIGQUERY_TABLE

Example command for DataflowRunner:

python -m gcp_variant_transforms.vcf_to_bq \
  --input_pattern gs://BUCKET/*.vcf \
  --output_table GOOGLE_CLOUD_PROJECT:BIGQUERY_DATASET.BIGQUERY_TABLE \
  --project "${GOOGLE_CLOUD_PROJECT}" \
  --temp_location gs://BUCKET/temp \
  --job_name vcf-to-bigquery \
  --setup_file ./setup.py \
  --runner DataflowRunner

Running VCF files preprocessor

The VCF files preprocessor is used for validating the datasets such that the inconsistencies can be easily identified. It can be used as a standalone validator to check the validity of the VCF files, or as a helper tool for VCF to BigQuery pipeline. Please refer to VCF files preprocessor for more details.

Running BigQuery to VCF

The BigQuery to VCF pipeline is used to export variants in BigQuery to one VCF file. Please refer to BigQuery to VCF pipeline for more details.

Running jobs in a particular region/zone

You may need to constrain Cloud Dataflow job processing to a specific geographic region in support of your project’s security and compliance needs. See Setting zone/region doc.

Additional topics

• Understanding the BigQuery Variants Table Schema
• Loading multiple files
• Variant merging
• Handling large inputs
• Appending data to existing tables
• Variant Annotation
• Partitioning
• Flattening the BigQuery table
• Troubleshooting

Development

• Development Guide
• Release process