Limits...
"Missing" G x E Variation Controls Flowering Time in Arabidopsis thaliana.

Sasaki E, Zhang P, Atwell S, Meng D, Nordborg M - PLoS Genet. (2015)

Bottom Line: The SNP-based scan identified several variants that had common effects in both environments, but found no trace of G x E effects, whereas the scan using local variance components found both.Furthermore, the G x E effects appears to be concentrated in a small fraction of the genome (0.5%).Our conclusion is that G x E effects in this study are mostly due to large numbers of allele or haplotypes at a small number of loci, many of which correspond to previously identified flowering time genes.

View Article: PubMed Central - PubMed

Affiliation: Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter (VBC), Vienna, Austria.

ABSTRACT
Understanding how genetic variation interacts with the environment is essential for understanding adaptation. In particular, the life cycle of plants is tightly coordinated with local environmental signals through complex interactions with the genetic variation (G x E). The mechanistic basis for G x E is almost completely unknown. We collected flowering time data for 173 natural inbred lines of Arabidopsis thaliana from Sweden under two growth temperatures (10°C and 16°C), and observed massive G x E variation. To identify the genetic polymorphisms underlying this variation, we conducted genome-wide scans using both SNPs and local variance components. The SNP-based scan identified several variants that had common effects in both environments, but found no trace of G x E effects, whereas the scan using local variance components found both. Furthermore, the G x E effects appears to be concentrated in a small fraction of the genome (0.5%). Our conclusion is that G x E effects in this study are mostly due to large numbers of allele or haplotypes at a small number of loci, many of which correspond to previously identified flowering time genes.

No MeSH data available.


Enrichment for a priori flowering time candidates in using local variance component analysis.Top row: enrichment and FDR (upper bound among a priori candidates). The horizontal dashed lines at 1 corresponds to no enrichment. Bottom row: Quantile-quantile plots comparing the distribution of likelihood ratios in windows containing a priori candidates with windows that do not. The different curves shows result after removing windows significant using more stringent thresholds (see text). The shaded region corresponds to a 95% confidence interval. See Methods for details.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4608753&req=5

pgen.1005597.g004: Enrichment for a priori flowering time candidates in using local variance component analysis.Top row: enrichment and FDR (upper bound among a priori candidates). The horizontal dashed lines at 1 corresponds to no enrichment. Bottom row: Quantile-quantile plots comparing the distribution of likelihood ratios in windows containing a priori candidates with windows that do not. The different curves shows result after removing windows significant using more stringent thresholds (see text). The shaded region corresponds to a 95% confidence interval. See Methods for details.

Mentions: Result for the full local and the common local effect tests were strongly correlated with their corresponding GWAS results (presented above), especially for genes with reasonably strong association with flowering, while GSNP x E and Glocal x E showed much lower correlation (S5 Fig). Because the variance component likelihood ratios are not calibrated, it is difficult to say whether any particular effect is significant. However, we can assess this using overrepresentation of a priori candidates as for MTMM above. In all tests (full local, common local and Glocal x E), a significant enrichment of a priori candidates exist for likelihood ratios of 5 or higher, for which FDR is less than 20% (Fig 4). Notably, this effect was observed for the Glocal x E effect test as well, whereas GSNP x E showed no evidence of overrepresentation (Fig 3). Thus the variance component analysis appears to capture G x E effects not captured by the marginal SNP GWAS.


"Missing" G x E Variation Controls Flowering Time in Arabidopsis thaliana.

Sasaki E, Zhang P, Atwell S, Meng D, Nordborg M - PLoS Genet. (2015)

Enrichment for a priori flowering time candidates in using local variance component analysis.Top row: enrichment and FDR (upper bound among a priori candidates). The horizontal dashed lines at 1 corresponds to no enrichment. Bottom row: Quantile-quantile plots comparing the distribution of likelihood ratios in windows containing a priori candidates with windows that do not. The different curves shows result after removing windows significant using more stringent thresholds (see text). The shaded region corresponds to a 95% confidence interval. See Methods for details.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4608753&req=5

pgen.1005597.g004: Enrichment for a priori flowering time candidates in using local variance component analysis.Top row: enrichment and FDR (upper bound among a priori candidates). The horizontal dashed lines at 1 corresponds to no enrichment. Bottom row: Quantile-quantile plots comparing the distribution of likelihood ratios in windows containing a priori candidates with windows that do not. The different curves shows result after removing windows significant using more stringent thresholds (see text). The shaded region corresponds to a 95% confidence interval. See Methods for details.
Mentions: Result for the full local and the common local effect tests were strongly correlated with their corresponding GWAS results (presented above), especially for genes with reasonably strong association with flowering, while GSNP x E and Glocal x E showed much lower correlation (S5 Fig). Because the variance component likelihood ratios are not calibrated, it is difficult to say whether any particular effect is significant. However, we can assess this using overrepresentation of a priori candidates as for MTMM above. In all tests (full local, common local and Glocal x E), a significant enrichment of a priori candidates exist for likelihood ratios of 5 or higher, for which FDR is less than 20% (Fig 4). Notably, this effect was observed for the Glocal x E effect test as well, whereas GSNP x E showed no evidence of overrepresentation (Fig 3). Thus the variance component analysis appears to capture G x E effects not captured by the marginal SNP GWAS.

Bottom Line: The SNP-based scan identified several variants that had common effects in both environments, but found no trace of G x E effects, whereas the scan using local variance components found both.Furthermore, the G x E effects appears to be concentrated in a small fraction of the genome (0.5%).Our conclusion is that G x E effects in this study are mostly due to large numbers of allele or haplotypes at a small number of loci, many of which correspond to previously identified flowering time genes.

View Article: PubMed Central - PubMed

Affiliation: Gregor Mendel Institute, Austrian Academy of Sciences, Vienna Biocenter (VBC), Vienna, Austria.

ABSTRACT
Understanding how genetic variation interacts with the environment is essential for understanding adaptation. In particular, the life cycle of plants is tightly coordinated with local environmental signals through complex interactions with the genetic variation (G x E). The mechanistic basis for G x E is almost completely unknown. We collected flowering time data for 173 natural inbred lines of Arabidopsis thaliana from Sweden under two growth temperatures (10°C and 16°C), and observed massive G x E variation. To identify the genetic polymorphisms underlying this variation, we conducted genome-wide scans using both SNPs and local variance components. The SNP-based scan identified several variants that had common effects in both environments, but found no trace of G x E effects, whereas the scan using local variance components found both. Furthermore, the G x E effects appears to be concentrated in a small fraction of the genome (0.5%). Our conclusion is that G x E effects in this study are mostly due to large numbers of allele or haplotypes at a small number of loci, many of which correspond to previously identified flowering time genes.

No MeSH data available.