Hi Jorge (and everybody),
this is a general question not specifically related to NGSEP. Hope it is appropriate here.
I am dealing with a set of GBS data (that I already analyzed by NGSEP) of very few plant accessions of the Camelina sativa species that features some differences for agronomic traits. I am wondering whether it is possible to try inferring associations between GBS data and phenotypic traits with any given software or tool given the very limited number of plants under analysis. For example by the analysis of VCF data in candidate genes or in the proximity of such genes. In fact, GBS is a complexity reduction way to investigate genomes but it doesn not, in principle, detects any variation occurring on genes. I do not know if there are ways to try interrogate GBS variants (when comparing two or more individuals) on selected genes or pathways.
Any comment or hint is welcome.
Thanks a lot
Filippo
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The standard way to make genotype / phenotype associations would be GWAS. However, GWAS requires more than 100 accessions, hopefully without a strong population structure. The underlying reason is that the most basic association analysis is a factorial ANOVA in which the possible genotypes are the categories. Hence, you need a good number of samples to have enough representation of each category. The lower the MAF, the larger number of samples you need to have some representation of genotypes with minor alleles and detect association.
You are right about the limitation of GBS on interrogation of gene variability. More in general, it is unlikely that causative SNVs for a trait can be identified with GBS data and, even if a causative SNV is covered, you still need a good sample size to detect a significant association.
The experimental design that you describe will help you to assess the variability of your samples. However, I am not aware of any way in which you can use GBS data of a low number of samples to detect genotype/phenotype associations. My apologies for the not so positive answer.
Best regards
Jorge
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
Anonymous
-
2022-06-20
Dear Jorge,
first of all thanks for your exhausive reply. So, you confirm the idea I had about the limitations (as far as I know) of GBS to infer genotype/phenotype associations from a reduced number of samples. Provided this, I could consider a resequencing approach of some of the Camelina plants with contrasting phenotypes followed by a targeted variant analysis.
Kind regards
Filippo
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
No problem. Just to clarify, low sample size is an issue for GWAS, even if you do WGS. Sequencing siblings with extreme phenotypes in a controlled population is a feasible alternative. You can as well increase sample size for some SNPs using any kind of tergeted genotyping.
Best regards
Jorge
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Jorge (and everybody),
this is a general question not specifically related to NGSEP. Hope it is appropriate here.
I am dealing with a set of GBS data (that I already analyzed by NGSEP) of very few plant accessions of the Camelina sativa species that features some differences for agronomic traits. I am wondering whether it is possible to try inferring associations between GBS data and phenotypic traits with any given software or tool given the very limited number of plants under analysis. For example by the analysis of VCF data in candidate genes or in the proximity of such genes. In fact, GBS is a complexity reduction way to investigate genomes but it doesn not, in principle, detects any variation occurring on genes. I do not know if there are ways to try interrogate GBS variants (when comparing two or more individuals) on selected genes or pathways.
Any comment or hint is welcome.
Thanks a lot
Filippo
Hi Filippo
The standard way to make genotype / phenotype associations would be GWAS. However, GWAS requires more than 100 accessions, hopefully without a strong population structure. The underlying reason is that the most basic association analysis is a factorial ANOVA in which the possible genotypes are the categories. Hence, you need a good number of samples to have enough representation of each category. The lower the MAF, the larger number of samples you need to have some representation of genotypes with minor alleles and detect association.
You are right about the limitation of GBS on interrogation of gene variability. More in general, it is unlikely that causative SNVs for a trait can be identified with GBS data and, even if a causative SNV is covered, you still need a good sample size to detect a significant association.
The experimental design that you describe will help you to assess the variability of your samples. However, I am not aware of any way in which you can use GBS data of a low number of samples to detect genotype/phenotype associations. My apologies for the not so positive answer.
Best regards
Jorge
Dear Jorge,
first of all thanks for your exhausive reply. So, you confirm the idea I had about the limitations (as far as I know) of GBS to infer genotype/phenotype associations from a reduced number of samples. Provided this, I could consider a resequencing approach of some of the Camelina plants with contrasting phenotypes followed by a targeted variant analysis.
Kind regards
Filippo
Hi Filippo
No problem. Just to clarify, low sample size is an issue for GWAS, even if you do WGS. Sequencing siblings with extreme phenotypes in a controlled population is a feasible alternative. You can as well increase sample size for some SNPs using any kind of tergeted genotyping.
Best regards
Jorge