Output of METAL meta-analysis: which allele does the beta refer to?
1
0
Entering edit mode
7.6 years ago
smnlalli • 0

Hi,

I have run a meta-analysis using METAL, and I have an output file (ending in ".tbl") including an "Effect" column and an "Allele1" and "Allele2". I would like to compare the effect column of this analysis to another dataset, but in order to do so I need to align the betas depending on which allele the effects refer to. I was wondering if I can assume that the effect is computed for the allele specified in the "Allele1" field.

The Wiki documentation for METAL talks about the first study included in the meta-analysis being used as the reference for the alleles, so if I am understanding this correctly the "Effect" is expressed in reference to "Allele1". If this is not the case, how can I find the allele that is used as the effect allele?

Thank you very much for all your help,

Simone

METAL meta-analysis • 6.3k views
0
Entering edit mode

What's the answer? I think I met the same question.

1
Entering edit mode
2.2 years ago
Olliepain ▴ 10

METAL creates a .tbl.info file when running the analysis which describes the output. It states that the Effect column corresponds to the "overall estimated effect size for allele1".

0
Entering edit mode

Hi,

Can you please tell me how to include BETA and SE in the output of METAL?

When I just run:

./metal < metal.txt


I just get this columns:

head  METAANALYSIS1.TBL

MarkerName  Allele1 Allele2 Weight  Zscore  P-value Direction
rs2326918   a   g   19586.00    0.254   0.7991  --+
rs668796    c   g   19586.00    -2.064  0.03904 -+-

vi metal.txt

MARKER   SNP
DEFAULTWEIGHT   1211
ALLELE   A1 A2
EFFECT   BETA
STDERR   SE
PVAL     P

PROCESS EDIC.GWAS.reduce.txt

MARKER   SNP
ALLELE   A1 A2
DEFAULTWEIGHT   1363
EFFECT   BETA
STDERR   SE
PVAL     P

PROCESS GOKIND.GWAS.reduce.txt

OUTFILE EDIC_GOKIND_UKB_META.tbl
ANALYZE

0
Entering edit mode

I guess if you set your analysis scheme as "STDERR" it will give you BETA and SE values.

SCHEME STDERR - classical approach, uses effect size estimates and standard errors