Phasing pipeline

This workflow is used to generate a phased VCF by using Beagle/Shapeit and emulates the analyses from the [ phase 3] of the 1000 genomes project. It can be run after generating and integrated call set by using the PyHive::PipeConfig::PHASING.pm

This pipeline is designed to be run for SNPs or INDELs independently or for both variant types together in the same VCF. Important: This workflow can only analyze biallelic variants and it will crash if you try to analyze multiallelic sites.

1. Input preparation:

This pipeline will take as input the VCF file that contains either SNPs, INDELs or both types together. In order to generate a combined VCF containing both SNPs+INDELs from 1 SNP VCF + 1 INDEL VCF you can do the following:

bcftools concat input1.snps.vcf.gz input2.indels.vcf.gz -o combined.snps_indels.vcf.gz -Oz -a

2. Preparing the environment

Modify your $PYTHONPATH to include the required libraries:

export PYTHONPATH=${ehive_dir}/wrappers/python3/:$PYTHONPATH

Modify your $PERL5LIB to include the required libraries:

export PERL5LIB=${ehive_dir}/modules/:${igsr_analysis_dir}/:${PERL5LIB}

Modify your $PATH to include the location of the eHive scripts:

export PATH=${ehive_dir}/scripts/:${PATH}

• Install dependency

  1. Clone repo by doing git clone https://github.com/igsr/igsr_analysis.git in the desired folder
  2. pip install ${igsr_analysis_dir}/dist/igsr_analysis-0.91.dev0.tar.gz
  3. Modify $PYTHONPATH to add the folder where your pip installs the Python packages

  And you are ready to go!

• Conventions used in this section

  • ${igsr_analysis_dir} is the folder where you have cloned https://github.com/igsr/igsr_analysis.git
  • ${ehive_dir} is the folder where you have cloned https://github.com/Ensembl/ensembl-hive.git

3. Databases

The pipeline uses two databases. They may be on different servers or the same server.

3.1 The ReseqTrack database

The pipeline queries a ReseqTrack database to find the VCF that will be filtered by the pipeline. It will also add file metadata for the final filtered VCF.

In order to create a ReseqTrack database use the following

commands:

mysql -h <hostname> -P <portnumber> -u <username> -p???? -e "create database testreseqtrack" # where testreseqtrack
                                                                                             # is the name you want
                                                                                             # to give to the ReseqTrack DB
mysql -h <hostname> -P <portnumber> -u <username> -p???? testreseqtrack < $RESEQTRACK/sql/table.sql
mysql -h <hostname> -P <portnumber> -u <username> -p???? testreseqtrack < $RESEQTRACK/sql/views.sql

• Conventions used in this section:

$RESEQTRACK is the folder where you have cloned https://github.com/EMBL-EBI-GCA/reseqtrack.git

3.2 The Hive database

This is database is used by the Hive code to manage the pipeline and job submission etc. The pipeline will be created automatically when you run the init_pipeline.pl script. Write access is needed to this database.

4. Initialise the pipeline

The pipeline is initialised with the hive script init_pipeline.pl. Here is an example of how to initialise a pipeline:

init_pipeline.pl PyHive::PipeConfig::INTEGRATION::PHASING \
                 -pipeline_url mysql://g1krw:$DB_PASS@mysql-rs-1kg-prod:4175/hive_dbname \
                 -db testreseqtrack \
                 -pwd $DB_PASS \
                 -hive_force_init 1

The first argument is the the module that defines this pipeline. Then -pipeline_url controls the Hive database connection details, in this example:

g1krw= username
$DB_PASS= password
mysql-rs-1kg-prod= hostname
4175= Port number
hive_dbname= Hive DB name

Then -db is the name of the Reseqtrack database name used in the section 2.1 -pwd is the ReseqTrack DB password

The rest of the options are documented in the PyHive::PipeConfig::INTEGRATION::PHASING.pm module file. You will probably want to override the defaults for many of these options so take a look.

5. Seeding the pipeline

In order to seed the pipeline with the VCF file that will be analyzed use the hive script seed_pipeline.pl:

seed_pipeline.pl \
                 -url mysql://g1krw:$DB_PASS@mysql-rs-1kg-prod:4175/hive_dbname \
                 -logic_name find_files \
                 -input_id "{ 'file' => '/path/to/file/input_file.txt' }"

Where -url controls the Hive database connection details and /path/to/file/input_file.txt contains the filename of the VCF to be analyzed.

6. Sync the hive database

This should always be done before [re]starting a pipeline:

Run e.g.:

beekeeper.pl -url mysql://g1krw:{password}@mysql-g1k:4175/my_hive_db_name -sync

where -url are the details of your hive database. Look at the output from init_pipeline.pl to see what your url is.

7. Run the pipeline

Run e.g.:

beekeeper.pl -url mysql://g1krw:{password}@mysql-g1k:4175/my_hive_db_name -loop &

Note the ‘&’ makes it run in the background.

Look at the pod for beekeeper.pl to see the various options. E.g. you might want to use the -hive_log_dir flag so that all output/error gets recorded in files.

While the pipeline is running, you can check the ‘progress’ view of the hive database to see the current status. If a job has failed, check the msg view.