|
Post by kzar0004 on Aug 13, 2016 12:14:22 GMT
Dear All,
I am new to GCTA and have been trying to use the program to estimate the variance in my phenotypes (e.g. BMI, IQ scores from various cognitive tests) from n=202 subjects explained by my SNPs ( approximately 571074). My data is in the Binary PED files (plink.bed, plink.bim, and plink.fam).
I started off with this command:
gcta_mac --bfile Danes181_NoMind --autosome --make-grm --out Danes181_NoMind
And then tried to do REML analysis using the GRM from the previous command above using this command:
gcta_mac --grm Danes181_NoMind --pheno Danes197.Noheader --reml --out Danes181_NoMind
But then I receive this output
Reading IDs of the genetic relationship matrix (GRM) from [Danes181_NoMind.grm.id]. 181 IDs read from [Danes181_NoMind.grm.id]. Reading the GRM from [Danes181_NoMind.grm.gz]. Pairwise genetic relationships between 181 individuals are included from [Danes181_NoMind.grm.gz].
Error: can not open the file [Danes197.Noheader] to read.
My phenotype file has no header, is a text file, and the columns are: family id, individual id, phenotype...
What am I doing incorrect?
I look forward to your response.
Best wishes,
K
|
|
|
Post by Jian Yang on Aug 15, 2016 12:07:15 GMT
It has nothing to do with the format of the file. The error message says that the program couldn't find the file in the directory.
|
|
|
Post by kzar0004 on Aug 15, 2016 18:28:39 GMT
Thank you for your reply. The file is in the correct directory..? I am not sure if that was the problem but it seems to be fine now.
I have a question regarding my REML output. In the output below the phenotype I used was results from a cognitive examination for n=162 subjects.
Source Variance SE V(G) 52.372958 26.878553 V(e) 34.147764 24.054993 Vp 86.520722 9.958276 V(G)/Vp 0.605323 0.285860 logL -439.013 logL0 -440.225 LRT 2.422 df 1 Pval 0.06 n 162
Is it correct to interpret from the above that 60% of the variance observed in the cognitive scores is explained by all the SNPs?
Furthermore I am slightly confused by how to understand the "Pval 0.06". In the GCTA homepage under section: GCTA-GREML: estimation of the phenotypic variance explained by the SNPs, point number '5' , the explanation of the P val is missing... How can the genetic variation explain 60% of the variation seen in the phenotype (cognitive exam) but this not yield a significant P val?
|
|
|
Post by kzar0004 on Aug 15, 2016 18:29:55 GMT
Also, in your homepage you write "If there are multiple GRMs included in the REML analysis, there will be multiple rows for the genetic variance ..."
Why would there be multiple GRMs?
|
|