Update released for the v7 Genomic Extraction Tool

The All of Us Researcher Workbench team has released an update to the CDRv7 genomic extraction tool to address an issue that may have resulted in incomplete output. The issue stemmed from a timing conflict, where genomic extraction results were occasionally generated before the required VCF data files were fully prepared.

Background

In the All of Us Researcher Workbench, you are able to utilize our point and click tool, the ‘Genomic Extraction Tool’, to extract variant data from the srWGS genomics dataset and save it as VCF (Variant Call Format) file for export to a Jupyter Notebook for analysis using Hail, PLINK, or other analysis tools of your choosing.

The v7 genomics extraction tool erroneously excluded some variants, meaning you may not have received all the variants for all participants in your final VCF files. Our team noted inconsistencies in the code operations for the tool, specifically where two main operations required to generate a VCF file weren’t properly coordinated. This could result in two extractions of the same cohort producing different results, which we noticed through inconsistent Variant Allele Frequencies (VAFs). The VAF is the percentage of sequencing data that supports a specific genetic variant. If the VAFs differ for the same cohort, the extraction was inconsistent.

Impact and Resolution

As a result of this issue, variant annotations (such as allele frequencies) generated using the VCFs generated from the extraction tool may be incorrect because they did not take the full cohort into consideration. It's important to note that the issue did not introduce false positives; variants were only excluded, not incorrectly added.

The issue has now been resolved in the updated CDRv7 genomic extraction tool. Please also note that the CDRv8 genomic extraction tool was not affected by this issue.

Although spurious results are unlikely, if you used the v7 genomics extraction tool to generate files prior to October 7th, 2025, you may consider reviewing your analysis results to assess the robustness of your findings. 

If you have any questions regarding this update, please contact us at support@researchallofus.org.