Hi, we are working with imputed data. The workflow is that we select specific region of SNPs on imputed data and test the associations between SNPs and phenotype. After QC (hwe, MAF, INFO), 3000+ SNPs were selected. My question is : can we use BOLT LMM to do the association testing? Moreover, how to apply BOLT LMM when there are only 3000+ SNPs. Besides, when we uses plink to merge two files, the warning shows: some variants has with 3+ alleles present, how can we deal with this problem?
These questions have troubled us for a long time. Thank you very much if you can give us some advice.
The BOLT-LMM is installed within swiss-army-knife. Although, it's not the latest version, but it is relatively new, so you should be able to use it for your research work.
Which impute data you are using? Your number of variants after filtering sounds too little to me.
Thank you for your help, our imputed data is from bulk/imputation/UKB imputation from genotype. We only focus on 2 genes which is on chr3 and chr10. After QC (maf 0.01 hwe 10e-6 INFO 0.8), 3000+variants are remained. We wonder to know if it is suitable to uses BOLT LMM or we could use simple linear regression in such tittle SNPs.
Something also went wrong: Intel MKL ERROR: Parameter 6 was incorrect on entry to DGELS. and/home/dnanexus/job-GV3jY48J92XfKQ6px4fggv3v.code.sh: line 69: 7987 Segmentation fault
We have trapped here for a long time. Thank you very much for your kindly help.
I see. That makes sense. I just want to make sure there isn't something wrong with data.
For BOLT-LMM usage, I'm afraid it would be best to ask BOLT-LMM developer. I know how to get around this with other tools, but I think it's best for you to check if it's possible to do with BOLT-LMM first.
?Thank you for your help. Besides, we found no imputed data of TOPMED and GEL in Bulk/Imputation. Are theses available now? We found they are retained in our basket.
Yes, it should be available now. For now, please dispense a new project and copy data over rather than refresh project. The refresh queue has some issue, and DNAnexus engineering team are trying to fix it.
Also, please see the post about issue with TOPMed and GEL data discussed in this thread.
?Thank you for your kindly help. We have tried to dispense a new project and copy the folder /Bulk/Imputation to the new project, but we found no imputed data of TOPMED and GEL. We wonder to know if the method we used is wrong.
Comments
8 comments
The BOLT-LMM is installed within swiss-army-knife. Although, it's not the latest version, but it is relatively new, so you should be able to use it for your research work.
Which impute data you are using? Your number of variants after filtering sounds too little to me.
Thank you for your help, our imputed data is from bulk/imputation/UKB imputation from genotype. We only focus on 2 genes which is on chr3 and chr10. After QC (maf 0.01 hwe 10e-6 INFO 0.8), 3000+variants are remained. We wonder to know if it is suitable to uses BOLT LMM or we could use simple linear regression in such tittle SNPs.
Best wishes
Yimeng
The code we used in Swiss Army Knife is as following:
warnings for step 3 is: Error: Line 2 of chr10.pvar has fewer tokens than expected.
we have tried to run BOLT LMM for each chr using the code:
bolt --fam chr10.fam --bim chr10.bim --bed chr10.bed --phenoFile pheno.txt --phenoCol ra --covarFile cov.txt --covarCol array --covarCol sex --covarCol age_g1 --covarMaxLevels 200 --lmm --numLeaveOutChunks --LDscoresFile LDSCORE.1000G_EUR.tab.gz --LDscoresMatchBp --statsFile output
Something also went wrong: Intel MKL ERROR: Parameter 6 was incorrect on entry to DGELS. and /home/dnanexus/job-GV3jY48J92XfKQ6px4fggv3v.code.sh: line 69: 7987 Segmentation fault
We have trapped here for a long time. Thank you very much for your kindly help.
I see. That makes sense. I just want to make sure there isn't something wrong with data.
For BOLT-LMM usage, I'm afraid it would be best to ask BOLT-LMM developer. I know how to get around this with other tools, but I think it's best for you to check if it's possible to do with BOLT-LMM first.
?Thank you for your help. Besides, we found no imputed data of TOPMED and GEL in Bulk/Imputation. Are theses available now? We found they are retained in our basket.
Yes, it should be available now. For now, please dispense a new project and copy data over rather than refresh project. The refresh queue has some issue, and DNAnexus engineering team are trying to fix it.
Also, please see the post about issue with TOPMed and GEL data discussed in this thread.
https://community.dnanexus.com/s/question/0D5t000004SBxtyCAD/potential-issues-with-imputed-data
?Thank you for your kindly help. We have tried to dispense a new project and copy the folder /Bulk/Imputation to the new project, but we found no imputed data of TOPMED and GEL. We wonder to know if the method we used is wrong.
Best wishes,
Yimeng
Please sign in to leave a comment.