MultiXcan output not matching

[david davtian]

Hi everyone, 

I have a quick question about my MultiXcan output. It ran smoothly with no issues, but the pvalues/zscores in the output for best/worst tissues are not matching the corresponding values for each gene in their respective single tissue SPrediXcan results.

Note, I used the latest inflation corrected models and parameters to run SPrediXcan. Could this be the reason why there is conflict between these values? If not do you know why this could be? 

Thank you very much for your help, 

Best regards, 

David

[david davtian]

I forgot, this is the command I used to run MultiXcan : 

./SMulTiXcan.py \
--models_folder ~/Endometriosis_TWAS/GTEx_models/ \
--models_name_pattern "en_(?P<tissue>.*)\.db" \
--snp_covariance ~/gtex_v8_expression_elastic_net_snp_smultixcan_covariance.txt.gz \
--metaxcan_folder ~/Endometriosis_TWAS/TWAS_results/ \
--metaxcan_file_name_parse_pattern "SPrediXcan_(?P<tissue>.*)\.csv" \
--gwas_file ~/Endometriosis_TWAS/GWAS/final_META_endometriosis_EUR2.txt.gz \
--snp_column MarkerName \
--effect_allele_column Allele1 \
--non_effect_allele_column Allele2 \
--zscore_column Zscore \
--model_db_snp_key rsid \
--keep_non_rsid \
--cutoff_condition_number 30 \
--verbosity 9 \
--output ~/Endometriosis_TWAS/res_MultiXcan.csv

[Hae Kyung Im]

Could you share a portion of the output?

--
You received this message because you are subscribed to the Google Groups "PrediXcan/MetaXcan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to predixcanmetax...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/predixcanmetaxcan/9e92b42e-d869-4bd2-b47b-ac5ffaaf80e6n%40googlegroups.com.

[david davtian]

Sure 

For example this is my MultiXcan output. 

And if I look into this first PNKP gene in Breast_Mammary SPrediXcan results, this is what I have

Neither the pvalue or even the uncalibrated_pvalue are matching.