ERROR: No 'SNP' column found

[mehrnooshshokouhi]

Hi, please have a look at my config. file bellow:
SCHEME STDERR
GENOMICCONTROL ON
AVERAGEFREQ ON
MINMAXFREQ ON
VERBOSE OFF
CUSTOMVARIABLE TotalSampleSize
LABEL TotalSampleSize as N

SEPARATOR TAB
MARKER SNP
ALLELE EffectAllele OtherAllele
FREQ MAF
EFFECT Beta
STDERR SE
WEIGHT N
PVALUE Pvalue
PROCESS /DATA/users/m.shokouhi/projects/GWAS-for-CBC/metal/meta_analysis/icogs.txt
PROCESS /DATA/users/m.shokouhi/projects/GWAS-for-CBC/metal/meta_analysis/onco.txt

OUTFILE /DATA/users/m.shokouhi/projects/GWAS-for-CBC/metal/meta_analysis/icogs_onco_meta. tbl
ANALYZE HETEROGENEITY
QUIT

Also, please have a look at the collumn name of my input data bellow:

colnames(icogs)
[1] "SNP" "CHR" "BP" "EffectAllele" "OtherAllele" "SE" "Pvalue" "N" "Beta" "Z_STAT"
[11] "MAF"
colnames(onco)
[1] "SNP" "CHR" "BP" "EffectAllele" "OtherAllele" "SE" "Pvalue" "N" "Beta" "Z_STAT"
[11] "MAF"

Why I still have the error of No 'SNP' column found?!

Looking forward to hearing from you.

[mehrnooshshokouhi]

Hi again,
I have another issue as well.
I realized if I write the name of column within "" (for example "SNP"), then the meta analysis would work.
But I am not sure if it works correctly.
This is because I found that the first and second alleles swap in many SNPs!
I do not know how to fix it.
I would appreciate any guide.
Cheers,
Mehrnoosh

[welchr]

When writing out icogs.txt, it may be that R quoted the column names, which is why you need to then keep the quotations in the METAL config file. If instead you do write.table(..., quote=F) then R will not keep the quotation marks.

In regards to allele swaps - METAL will keep track of which alleles were provided by each study and flip accordingly so that effect sizes are relative to the same reference allele. See also:

• https://genome.sph.umich.edu/wiki/METAL_Documentation#Approach:~:text=The%20z%2Dstatistic%20summarizes%20the%20magnitude%20and%20the%20direction%20of%20effect%20relative%20to%20the%20reference%20allele%20and%20all%20studies%20are%20aligned%20to%20the%20same%20reference%20allele
• https://genome.sph.umich.edu/wiki/METAL_Documentation#:~:text=We%20strongly%20recommend,resolve%20strand%20inconsistencies

[mehrnooshshokouhi]

Thank you for your prompt reply.
I am still confused about flipping alleles! Please check the table bellow. You see that the input data (two studies) consider the effect allele clearly. While in metal, it flipped the alleles and gave the beta for the reference allele, and of course showed the negative direction of the input studies for their reference allele. Additionally, it did not flip alleles for some other SNPs. What is the logic behind of this?
Best, Mehrnoosh

SNP/ Effect.a.x/ Other.a.x/ Beta.x/ Effect.a.y/ Other.a.y/ Beta.y/ Allele1/ Allele2/ Effect/ Direction

rs112149573:52581245:G:T / T / G / 0.149 / T / G / 0.129 / g / t / -0.1387 / --
rs12600239:52538900:C:T / T / C / 0.139 / T / C / 0.112 / c / t / -0.1247 / --
rs132289:29551872:A:G / G / A / 0.511 / G / A / 0.127 / a / g / -0.3547 / --

16:52582774:TAA:TA / TA / TAA / 0.137 / TA / TAA / 0.106 / ta / taa / 0.120 / ++
rs113333861:52614695:TA:T / T / TA / 0.166 / T / TA / 0.101 / t / ta / 0.130 / ++
rs132365:29599486:G:C / C / G / 0.598 / C / G / 0.096 / c / g / 0.395 / ++