SNP filters in GEMMA about gemma HOT 7 CLOSED

Problem solved by using -miss 1

GEMMA manual typo

'SNPs with missingness below 5% will not be included in the analysis' should be 'SNPs with missingness above 5% will not be included in the analysis'

from gemma.

Yes, that's a typo. You can just use "-maf 0" to include all SNPs (or more precisely, use "-miss 0 -maf 0 -r2 1 -hwe 0"). (1) "-notsnp" is not exactly the same as "-miss 0 -maf 0 -r2 1 -hwe 0" as it still check missingness and r2. (2) Even if you specify a list, some of them still could be filtered out.

from gemma.

Thanks!

from gemma.

Hi Xiang,
I am running Gemma 0.98.1 with these options:
-miss 0 -maf 0 -r2 1 -hwe 0 -snps my_snplist
In principle, all the variance in my input file should be analyzed, but I got the report looks like this:

## number of total SNPs/var        =    20829
## number of considered SNPS       =    20776
## number of analyzed SNPs         =     7695

So I think there is still filtering performed on the dataset. My questions are,

1. Could you specify what is the reason it does not analyze all the SNP with the above options? Is there any other default SNP filtering ?
2. And what has been filtered out during the total>considered>analyzed steps?

And one little suggestion, is it possible to specify due to which QC step the variance were filtered out in log file? e.g. 1000 SNPs were filtered out due to missing genotype, etc.

from gemma.

Hi,

was this solved?

I also have the same issue. Specifying flags for filtering such as
-miss 0
-maf 0
-r2 1
-hwe 0 \

I still miss ~half of the SNPS (from ~40million to ~20 million)

So it looks like there is some other sort of filtering performed?
I'm using Gemma_0.98.1

thanks in advance

from gemma.

Please use the mailing list.

from gemma.

good find, thanks!

from gemma.