Importing gnomAD Variant Dataset with Hail 0.2

I am trying to open the gnomAD Variant Dataset (VDS) with Hail 0.2:

import hail as hl
import hail.expr.aggregators as agg

mt = hl.read_matrix_table('gnomad.exomes.r2.0.2.sites.vds')

However I get the following error:

Hail version: devel-73b493d
Error summary: HailException: metadata does not contain file version: gnomad.exomes.r2.0.2.sites.vds/metadata.json.gz

Should I bite the bullet and uninstall Hail 0.2 and head back to 0.1? Or is there an easy fix for this in 0.2? Or is this error not related to Hail version, but the way I am trying to import the file or have set Hail up on my computer? I have no problem running the GWAS tutorial that comes with the developmental 0.2 version.

Thank you in advance for any help!

The file formats in Hail 0.1 and 0.2 are incompatible. I believe that all the gnomad files are from 0.1, but the gnomAD team is planning a release of only 0.2 files in the next few weeks.

so, to use the gnomad stuff, you’d need to do one of the following:

  1. Import from the VCFs using 0.2
  2. Use 0.1 to read the VDSes (I wouldn’t recommend this)
  3. Wait a few weeks

Thanks for the quick response! I will have a go with the VCFs and look forward to the new gnomAD releases.

How can I get notices about when the gnomAD VDS files for Hail 0.2 are to be available? Is there a web page or blog I should watch?

I expect they will be available on the gnomAD homepage by the time Konrad gives his ASHG talk tomorrow morning (Oct 17).

It’s live: