Thank you for this amazing tool!
I've run FOCUS to conduct fine-mapping of TWAS signals, and I've also run FUSION before that as well. I'm using the same reference panel (1000G), GWAS summary statistics, and eQTL weights (GTEx v8).
However, the marginal Z score from FOCUS differs quite a lot from the Z score from FUSION. Their correlation is only 0.01.
cor(focus$twas_z_pop1, fusion$TWAS.Z, use = "complete.obs")
[1] 0.01520323
It'd be great if you could help me look into what the reason for this discrepancy is. This is my script:
for i in {1..22}; do
Rscript ~/project/fusion/fusion_twas-master/FUSION.assoc_test.R \
--sumstats ~/project/data/ATH_model16_cleaned_snps_qpval.0.05.sumstats.gz \
--weights ~/project/fusion/fusion_twas-master/WEIGHTS/GTExv8.EUR.Adipose_Subcutaneous.pos \
--weights_dir ~/project/fusion/fusion_twas-master/WEIGHTS --GWASN 468954 \
--ref_ld_chr ~/project/fusion/fusion_twas-master/LDREF/1000G.EUR. --perm 10000 --perm_minp 0.05 \
--chr ${i} --out ~/project/sumstats/ATH/FUSION/gtex8/adipose_tissue_subcut/chr${i}.dat;
done
cut -f 2,3,4 ~/project/sumstats/ATH/FUSION/gtex8/adipose_tissue_subcut/chr*_joint.report | grep -v CHR > ~/project/sumstats/ATH/FUSION/gtex8/adipose_tissue_subcut/loci_chr1_2.txt
sed -i '1i\chrom\tstart\tstop' ~/project/sumstats/ATH/FUSION/gtex8/adipose_tissue_subcut/loci_chr1_2.txt
focus import ~/project/fusion/fusion_twas-master/WEIGHTS/GTExv8.EUR.Adipose_Subcutaneous.pos fusion --tissue Adipose_Subcutaneous --name EUR --assay array --output fusion.gtexv8.adipose_tissue_subcut
focus finemap --locations ~/project/sumstats/ATH/FUSION/gtex8/adipose_tissue_subcut_newrun/loci_chr1_2.txt \
--trait ath_model16_at_subcut_gtex8 \
~/project/data/ATH_model16_cleaned_snps_qpval.0.05.sumstats.gz \
~/project/data/plink_LD_ref_panel/1000G_EUR_Phase3_plink_merged \
fusion.gtexv8.adipose_tissue_subcut.db \
--out ~/project/sumstats/ATH/FUSION/gtex8/adipose_tissue_subcut_newrun/focus_allchr
Thank you very much in advance.
Best regards,
Thanh Le
Thank you for this amazing tool!
I've run FOCUS to conduct fine-mapping of TWAS signals, and I've also run FUSION before that as well. I'm using the same reference panel (1000G), GWAS summary statistics, and eQTL weights (GTEx v8).
However, the marginal Z score from FOCUS differs quite a lot from the Z score from FUSION. Their correlation is only 0.01.
It'd be great if you could help me look into what the reason for this discrepancy is. This is my script:
Thank you very much in advance.
Best regards,
Thanh Le