Does it make sense to use UTMOST with a tissue that is not yet part of GTEX

[montenegrina]

Hello,

I would like to do TWAS using Retina tissue that is not yet part of GTEX and my UK Biobank GWAS data. Does using UTMOST in that case make sense and which series of steps I would need to perform in order to do that.

Thanks Ana

[Joker-Jerome]

Hi Ana,

Sure you could do that. The first thing you need is the imputation model for Retina tissue. After that,

1. You could perform single-tissue association test with the Retina imputation model; python2 ./single_tissue_association_test.py \ --model_db_path sample_data/weight_db_external/${tissue}.db \ --covariance sample_data/covariance_external/${tissue}.txt.gz \ --gwas_folder sample_data/GWAS \ --gwas_file_pattern SNP_gwas_mc_merge_nogc.tbl.uniq \ --snp_column SNP \ --effect_allele_column A1 \ --non_effect_allele_column A2 \ --beta_column b \ --pvalue_column p \ --output_file sample_data/results/${tissue}.csv done

2. You could add the retina imputation model to our framework and perform a cross-tissue test; See "Incorporating external eQTL/sQTL datasets" part in the README page.

Thanks!

[montenegrina]

Hi,

Thank you so much for getting back to me. Can you please explain to me what you mean with Retina imputation model and how to obtain that? Also I am seeing that you have data base for weights.Do you have anywhere instructions on how to make that data base for my Retina tissue?

Also how do I make covariance_external file?

Thanks Ana

On Mon, Feb 17, 2020 at 8:12 PM Zhaolong(Jerome) Yu < notifications@github.com> wrote:

Hi Ana,

Sure you could do that. The first thing you need is the imputation model for Retina tissue. After that,

1.

You could perform single-tissue association test with the Retina imputation model; python2 ./single_tissue_association_test.py --model_db_path sample_data/weight_db_external/${tissue}.db --covariance sample_data/covariance_external/${tissue}.txt.gz --gwas_folder sample_data/GWAS --gwas_file_pattern SNP_gwas_mc_merge_nogc.tbl.uniq --snp_column SNP --effect_allele_column A1 --non_effect_allele_column A2 --beta_column b --pvalue_column p --output_file sample_data/results/${tissue}.csv done 2.

You could add the retina imputation model to our framework and perform a cross-tissue test; See "Incorporating external eQTL/sQTL datasets" part in the README page.

Thanks!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Joker-Jerome/UTMOST/issues/10?email_source=notifications&email_token=ACF3RTCNY2HHB3AFCV4VXXDRDM72JA5CNFSM4KWYLBT2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEMAJGHI#issuecomment-587240221, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACF3RTDVSGDDUFONUI6VYX3RDM72JANCNFSM4KWYLBTQ .

[Joker-Jerome]

Hi Ana,

No problem, the imputation model here means the model by which you could obtain the genetically regulated expression level with genetype data. You could train this the imputation model with the help of CTIMP: https://github.com/yiminghu/CTIMP or other TWAS imputation methods.

As for the covariance file, we have precalculate the covariance file for our imputation models. You could calculate the covariance with the function in "5.3. Calculate the joint tissue covariance".

Thanks, Jerome

[montenegrina]

Thank you so much. So for 5.3

mkdir covariance_GTEx_external ## path for saving new covariance matrix (could take ~25GB space) mkdir sample_data/weight_db_GTEx_external ## path for saving imputation models across different tissues cp sample_data/weight_db_GTEx/ sample_data/weight_db_GTEx_external/ cp sample_data/weight_db_external/ sample_data/weight_db_GTEx_external/ python2 ./joint_covariance.py \ --weight_db sample_data/weight_db_GTEx_external/ \ --input_folder sample_data/dosage/ \ --covariance_output sample_data/covariance_GTEx_external/

Can you please tell me what is the difference between weight_db_GTEx and weight_db_external? I will make weight_db_external I guess with CTIMP but what about weight_db_GTEx? Are weight_db_GTEx weights for all GTEX tissues and do I need that in my case?

sample_data/dosage/ would be dosage files in my case for Retina?

On Mon, Feb 17, 2020 at 8:25 PM Zhaolong(Jerome) Yu notifications@github.com wrote:

Hi Ana,

No problem, the imputation model here means the model by which you could obtain the genetically regulated expression level with genetype data. You could train this the imputation model with the help of CTIMP: https://github.com/yiminghu/CTIMP or other TWAS imputation methods.

As for the covariance file, we have precalculate the covariance file for our imputation models. You could calculate the covariance with the function in "5.3. Calculate the joint tissue covariance".

Thanks, Jerome

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.