Entering edit mode
3.7 years ago
Kumar
▴
170
Hi all,
I have a set of 82 VCFs files, I merged these all into one as a genotype file and a phenotype of five traits as data. I am looking to make a possible association of these data using Mixed linear model. I tried TASSEL in many ways with different parameters. However, I am not getting any potential significant SNPs after multiple correction using FDR and bonferroni.
Please suggest any other software/pipelines, which can be useful for GWAS study.
what is the relatedness of the 82 individuals, if any?
Yes, these are 82 individuals from a tree.
ok good. How are you representing this tree to TASSEL - as a kinship file?
Hi, I am new in this analysis. I have prepared a phenotype file and a genotype file for TASSEL. I have total 82 individuals in phenotype file with five data columns. I tried with factor and without factor to make association at TASSEL but not getting any significant SNPs after multiple correction. Please let me know how I can represent tree to TASSEL.
https://smcclatchy.github.io/mapping/08-calc-kinship/
Hi, I am uncertain, does it need to run TASSEL. My issue is not getting significant SNPs using TASSEL, not a kinship matrix.
As I understand it, the goal of using a mixed model is to control for population structure, so you can identify which SNPs, for instance, make an individual taller regardless if they occur from a family that is already tall, and not attribute height to some other SNPs that tall family shares, like those that confer curly hair. If you don't provide a kinship matrix, you will get more false positives and false negatives.
Yes, I am calculating kinship matrix at TASSEL. My way is using TASSEL is: uploading genotype and phenotype, filter the genotype, run the PCA of genotype, intersect the PCA+phenotype+genotype. Then, I generate kinship matrix of the genotype file. For mixed liner model I ran TASSEL with this kinship matrix and the file (PCA+phenotype+genotype).
Is there a known trait/genotype association that you can add as a positive control to make sure you are doing this correctly?
Sorry, I am uncertain. Could you please elaborate (known trait/genotype association?), how to add positive control? I just followed TASSEL tutorial for GWAS.
What organism are you working with?
It is a tree Sequoiadendron giganteum (giant sequoia: also known as giant redwood)
The point of a positive control is to have some reassurance that you are encoding things correctly. There must be some known penetrant SNPs that determine some phenotype like the number of branches. (I think the name TASSEL is a poorly hidden pun on the number of tassel branches in maize.)
If you expose local moisture as a phenotype it seems likely one of the loci (or one of the 26 SNPs) here will light up https://onlinelibrary.wiley.com/doi/full/10.1002/ece3.6716
Thank you for sharing the research paper. However, I am still not sure how to improve the way to get significant SNPs from my TASSEL analysis. I am wondering to go with GEMMA analysis also.
You may not be able to torture TASSEL enough to produce significant SNPs but the first step of debugging any analysis which requires weird formats is to make sure you prepared your input data properly. How would you know you did it correctly unless you have a positive control?
Yes, I agree. Could you indicate any sample input file for TASSEL where I can use positive control. Thank you!
The positive controls should be known SNPs in your selection of redwoods that correspond to something that distinguishes these trees from each other
Should I add the known SNPs in my input phenotype file for TASSEL or after getting MLM matrix from TASSEL. Sorry , if I am asking very common question.
I am not a TASSEL user but multiple testing correction is working on your P-values of individual tests. To perform multiple testing corrections or not is a debated topic. The goal of the mixed model analysis or any analysis is not to just provide significant results.