Help Regarding Conversion Of Imputed Data: Snptest Format To Plink

Question

[Joey]

I was given imputed files/ chromosome in SNPTEST format (GEN/SAMPLE)..i.e. a total of 44 files. I want to work with PLINK for one of the GWAS analyses. I don't know if PLINK can handle SNPTEST files nad so I tried to convert each of the pairs to MAP/PED using GTOOL and then converted them to binary plink files before merging them all together to create a trio of bed/bim/fam files.

Q1) Is this the best way to work with these types of data?

Q2) When I used GTOOL, I saw that the MAP file that were created were not in the right format. Under CHR, everything was listed as zero (a normal behavior for GTOOL conversion according to their webpage) and so I used a 'awk' command to change the entries of the CHR column to the corresponding chromosome# before converting it to binary PLINK files. Is this the best way to do this?

Q3) Also, for the PED file, FID and IID were all zeroes for every person which was very weird. I again put in a hack (using 'awk' again) to interchange the first two columns of the GEN file so that the PED file comes out right. Again, I don't know if there are better ways to deal with these issues.

Any suggestions or work around would be very helpful.

Thanks,

-Joey

Answer 1

[zx8754]

I am going through the same process, SNPTEST -> GTOOL -> awk/cut/paste -> PLINK. Creating a quick bash script that would take gen/sample files as argument could be a solution.

Also, plink can read dosage files - http://pngu.mgh.harvard.edu/~purcell/plink/dosage.shtml

plink \
--noweb \
--MyImpute2Output.gen \
  format=3 \
  dose=1 \
  skip0=1 \
  skip1=1 \
  noheader \
--MyFamFile.fam \
--assoc \
--out MyImpute2Output_assoc