Question: Quantitative trait analysis
1
gravatar for toyan.j.p
4.7 years ago by
toyan.j.p30
United States
toyan.j.p30 wrote:

I have couple of queries regarding QTL (Enzyme activity) association with genotypes  which I did and I am a novice in this field

First one is regarding the Plink output file generated using --qassoc command for quantitative trait for unphased genotypes. ( --qassoc gives allelic association (I presume)?). Is there a program by which i can check genotypic association with a quantitative trait?(there is genotypic means in plink but it doesn't give a Pvalue). I was able to get significant P values but was not able to find what is the significance of Beta, SE, R2  and T values. Would any of these parameters help to say the percentage of variance in the quantitative trait explained due to the presence of a particular variant? 

I have also generated sliding window haplotypes for association with quantitative trait(4 marker & 6 marker)? What is the basis for window size determination(I have a total of 64 markers spanning 22kb)? and what does BETA, R2, STAT and PS in the output format signify. Is there a method by which we can estimate the haplotypic means for quantitative trait for the associated haplotypes?

Would you pls help me with this queries

 

 

       

 

ADD COMMENTlink modified 4.6 years ago by Biostar ♦♦ 20 • written 4.7 years ago by toyan.j.p30
1

One note: generally for QTL analysis plink is not very good, especially slow and doesnt correct properly for multiple testing. Nowadays its more common to use programs like http://www.bios.unc.edu/research/genomic_software/Matrix_eQTL/ incredible fast, easy and proper FDR correction

ADD REPLYlink written 4.7 years ago by Floris Brenk890
1

I second this recommendation: plink is not designed to work with a very large number of different phenotypes.  The plink 1.9 documentation explicitly recommends Matrix eQTL for that case.

With that said; Floris, can you elaborate on the problem with plink's multiple testing correction?  If the v1.07 math was actually wrong, rather than just slow, I need to correct it.

ADD REPLYlink written 4.6 years ago by chrchang5235.6k

I'm not a statistics expert, but the calculations of FDR in plink in the eQTL analysis (as far as I know) are done per phenotype (so per expression value) and not genome wide. I strongly recommend using matrix eqtl for this where the FDR is designed for this purpose.
 

ADD REPLYlink written 4.6 years ago by Floris Brenk890

The --adjust values should be genome-wide.  But yes, PLINK never performs multiple-phenotype corrections.

ADD REPLYlink written 4.6 years ago by chrchang5235.6k

Hi, Just curious his original query is for quantitative phenotypes (enzymatic activity), and matrix eQTL is an application for gene expression data containing thousands of transcripts. Do u think it is still applicable. Also, can u point where in plink 1.9 documentation is recommended the use of Matrix eQTL.

ADD REPLYlink written 2.6 years ago by dgrover610

BETA, R2 are explained in the PLINK documentation for PLINK 1.7:

http://pngu.mgh.harvard.edu/~purcell/plink/anal.shtml#qt

The default model is allelic association, but other models are possible using --model:

http://pngu.mgh.harvard.edu/~purcell/plink/anal.shtml#model

 

 

ADD REPLYlink written 4.7 years ago by Ahill1.6k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1486 users visited in the last hour