Meta analysis using METAL

Question

[khn]

Hello,

I conducted a meta-analysis of GWASs using METAL. The output file shows one of the variants with a p-value of 5e-9. This variant was actually included in only one set of summary statistics used for the meta-analysis. However, when I checked the original statistical values, the p-value was 5e-6 in the summary statistics, which differs from the results of the meta-analysis by METAL. I suspect that the variant was not properly meta-analyzed since it was present in only one set of summary statistics.

FYI, my setting was as below

CHROMOSOMELABEL CHR
POSITIONLABEL BP
MARKER SNP
ALLELE ALT REF
EFFECT BETA
STDERR SE
PVALUE P
WEIGHT N
SCHEME STDERR
FREQLABEL AAF
TRACKPOSITIONS ON
AVERAGEFREQ ON
MINMAXFREQ ON
CUSTOMVARIABLE TotalSampleSize
LABEL TotalSampleSize as N

How could the p-value in METAL change (though beta and se are still the same) when the variant is included in only one set of summary statistics?