RAF_SE columns in the GWAS summary statistic

[sherryfan030]

Hi there,
When I try to compute Fst using GEAR, I found there is no RAF_SE columns in my GWAS summary statistics and in 1KG reference panel frq.file.
How can I get RAF_SE columns? Should I sample the GWAS SNPs matched to the 1KG reference samples?
I'm not really sure what is going on. Thank you in advance for your help!

[gc5k]

not sure what you are doing...
could you show me the command you were running?

[sherryfan030]

not sure what you are doing... could you show me the command you were running?

hi gc5k,
I should have been more specific.

I have been trying to compute the Fst values with each of three 1KG samples (CEU, FIN, and TSI, respectively) using my GWAS summary statistic：

In my statistic file：
CHR BP A1 A2 EAF Beta SE Pval N SNP
1 721290 C G 0.037 -0.019 0.026 4.8e-01 95022 rs12565286
1 736289 A T 0.14 -0.016 0.013 2.2e-01 144996 rs79010578
1 752242 A T 0.0043 0.034 0.18 8.5e-01 106979 rs12090487

In reference panel file（1kg_CEU_rsSNP.frq.txt，obtained from demonstration data （https://github.com/gc5k/GEAR/wiki/Fpc）：
CHR SNP A1 A2 EAF NCHROBS EAFSE BP
1 rs3131972 A G 0.1616 198 0.0261585530752352 752721
1 rs3131969 A G 0.1313 198 0.0240012952091584 754182
1 rs3131967 T C 0.1313 198 0.0240012952091584 754334
1 rs1048488 C T 0.1616 198 0.0261585530752352 760912

then I perform a command but get an error：

-$ java -jar -Xmn7168m -Xms7168m -Xmx7168m gear.jar sfst --meta-batch fst_batch.txt --qt-size fst_batch_N.txt --key SNP EAF EAFSE A1 A2 CHR BP Pval --top 3 --out 2Cohort_sfst

---

GEAR [GEnetic Analysis Repository]
(C) 2013 Guo-Bo Chen, Zhi-Xiang Zhu
Branch master
Commit f0390db
Built on Thu Dec 30 11:49:38 CST 2021 with uncommitted changes

---

[INFO] Subcommand: sfst
[INFO] Options in effect:
[INFO] --meta-batch fst_batch.txt
[INFO] --qt-size fst_batch_N.txt
[INFO] --key SNP EAF EAFSE A1 A2 CHR BP Pval
[INFO] --top 3
[INFO] --out 2Cohort_sfst
[INFO]
[INFO] Analysis started: Thu Jan 06 21:20:27 CST 2022
[INFO]
[INFO] Analysing summary statistics analysis for quantitative traits.

[INFO] 5 cohorts are remained for analysis.

[INFO] Cannot find the p value column in 1kg_CEU_rsSNP.frq.txt
[INFO] Removed 1307 loci due to incorrect se.
[INFO] Removed 1 locus due to bad a1 allele(s).
[INFO] No p values provided. Genomic control factor is set to 1.
[INFO] Genomic control factor (lambda_gc) is calculated from the provided beta and se: ▒▒
[INFO] Read 906306 (of 907614) summary statistics from '1kg_CEU_rsSNP.frq.txt.'

[INFO] Cannot find the p value column in 1kg_FIN_rsSNP.frq.txt
[INFO] Removed 3936 loci due to incorrect se.
[INFO] Removed 1 locus due to bad a1 allele(s).
[INFO] No p values provided. Genomic control factor is set to 1.
[INFO] Genomic control factor (lambda_gc) is calculated from the provided beta and se: ▒▒
[INFO] Read 903677 (of 907614) summary statistics from '1kg_FIN_rsSNP.frq.txt.'

[INFO] Cannot find the p value column in 1kg_TSI_rsSNP.frq.txt
[INFO] Removed 334 loci due to incorrect se.
[INFO] Removed 1 locus due to bad a1 allele(s).
[INFO] No p values provided. Genomic control factor is set to 1.
[INFO] Genomic control factor (lambda_gc) is calculated from the provided beta and se: ▒▒
[INFO] Read 907279 (of 907614) summary statistics from '1kg_TSI_rsSNP.frq.txt.'

[INFO] Cannot find the se value column in diabetes_Fst.txt
[INFO] GEAR quit.

---

I guess it's because there is no SE of EAF columns（sampling variance of EAF，as mentioned in https://github.com/gc5k/GEAR/wiki/Fpc） in my GWAS summary statistics.

Should I calculate the sampling variance of EAF in my summary statistic？
or could I perform this command without EAFSE columns included？

Thanks！

Lee

[gc5k]

It seems that you are engaged in meta-analysis.
You may write to me for more details here chenguobo@gmail.com.