Does anybody else encounter random read errors when reading imputed genotypes ?

Hi!   I am running scripts that read the imputed UKB data with bgenix using swissarmyknife.   On some occasions the job fails with:    ExecutionError: Detected slow performance in the worker's file system. Terminating  job-GXGYbFjJ8xY31k6vx1g1xkf3 with a non-billable ExecutionError.   On other occasion the job terminates, but throws an error when using bgenix:   Error: Invalid compressed SNP block in .bgen file.   Rerunning exactly the script makes the error "go away" (execution of the same dx run command, remotely, I am not touching anything else).   Interestingly this always happens when the job is working on the MHC region of Chr6, which is exceptionally dense in SNPs.   Is there an issue with the file system, maybe?   Thanks a lot for your thoughts,   Karsten.

Comments

5 comments

  • The latter error is a problem, as bgenix recovers and continues to work, but some SNPs just miss from the analysis.

     

    No I am screening my log files for this error to make sure everything is in order ;-)

    0
  • Comment author
    Ondrej Klempir DNAnexus Team

    Hi @Karsten Suhre?,

     

    thanks for sharing your observation with Community. Which field id you use? Is this post related to the previous discussion here? https://community.dnanexus.com/s/question/0D5t000004SBxtyCAD/potential-issues-with-imputed-data

     

    FYI, @Chai Fungtammasan? 

    0
  • It's not a data field, but the imputed genotype data in the bulk folder. I do something like this:

     

    ln -sf /mnt/project/Bulk/Imputation/UKB imputation from genotype/ impu

     

    bgenix -g impu/ukb22828_c6_b0_v3.bgen -incl-range 06:32105699-33105699

     

    plink2 --bgen region.bgen ref-first --sample impu/ukb22828_c6_b0_v3.sample --pheno pheno.txt --pheno-name nratio,nmet1,nmet2 --covar cova.txt --covar-name age,sex,genoPC1,genoPC2,genoPC3 --out nratio.6 --covar-variance-standardize --mac 100 --maf 0.01 --hwe 1e-15 --mind 0.1 --geno 0.1 --linear hide-covar --adjust

     

    which *sometimes* crashes with

     

    Error: Invalid compressed SNP block in .bgen file.

    0
  • Comment author
    Chai Fungtammasan DNAnexus Team

    I think this might be a bug with dxfuse. Have you contacted the ukbiobank-support@dnanexus.com? They might need to pass this to engineer team to investigate.

     

    0
  • Comment author
    Chai Fungtammasan DNAnexus Team

    I can create the ticket for you too, but it might be best for you to get in touch directly with them since they might need you to share the project with support team.

     

    0

Please sign in to leave a comment.