WORKFLOW

To generate the LD files for a given population (let's say African (afr)) do the following

Note, define population as the shortened form as seen in gnomAD website. African = afr, Non-finnish european = nfe, Estonian = est, etc...

1. Install Hail

Install the hail package which is leveraged to query and operate on the gnomAD blockmatrices downstream.

pip3 install hail

2. Update Global Hyperparameters

Navigate to hyperparams.py to update some hyper-parameters that the rest of scripts will use. Where this repository is git cloned,

  1. Update the global_directory_path variable such that the path is ".../gnomAD_LD_Easy_Querying/"

  2. Update the "sherlock_partition_string" to account for which partitions you will be using (specific to your lab). For example, I use "qsu,zihuai". One could also use "zihuai" etc. 

3. Download gnomAD files

Download from the gnomAD downloads page

1. For your population of choice, copy the "LD matrix Hail Block Matrix" file (ends in .bm) google URL. 

2. navigate terminal to /downloading_data_from_gnomAD/

3. Download file by running the following from command line:
    $python3 run_download_script.py <pasted_url_from_step1>

4. File will be downloaded in /gnomAD_downloaded_files/ repo

5. Repeat step 1 but, Copy the "Variant indices hail table" file (ends in .ht) google URL instead. You do not need to wait for the above script to finish - they can be running in parallel

4 Define Independent Blocks

First, wait for step 2 to finish. Now we need to define the independent blocks for our population of choice using the paper found here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4731402/ Moreover, we want to know which independent blocks each hit in our LD matrices belongs to. We thus create one file per chromosome, which contains a 2D list specifying the hits in each independent block (partition)

1. Navigate terminal to /generating_independent_partitions/ and run $python3 run_retrieve_independent_blocks.py nfe
   where 'nfe' is the population so could be 'afr', etc..

2. Step 1 should result in 22 files inside "gnomAD_LD_Easy_Querying/independent_partitions/nfe/

5 Generate LD Blocks

Now, we are ready to generate the LD matrices that will be saved inside gnomAD_LD_Easy_Querying/LD_matrices/nfe where nfe can be your population of choice consistent with the above

1. Navigate terminal to generating_LD_matrices/ and run:
$python3 run_populate_LD_matrices.py nfe, again where 'nfe' is changed to the population of choice

License

Distributed under the ASossin License

(back to top)