Integration of a systems biological network analysis and QTL results for biomass heterosis in Arabidopsis thaliana

Abstract

To contribute to a further insight into heterosis we applied an integrative analysis to a systems biological network approach and a quantitative genetics analysis towards biomass heterosis in early Arabidopsis thaliana development. The study was performed on the parental accessions C24 and Col-0 and the reciprocal crosses. In an over-representation analysis it was tested if the overlap between the resulting gene lists of the two approaches is significantly larger than expected by chance. Top ranked genes in the results list of the systems biological analysis were significantly over-represented in the heterotic QTL candidate regions for either hybrid as well as regarding mid-parent and best-parent heterosis. This suggests that not only a few but rather several genes that influence biomass heterosis are located within each heterotic QTL region. Furthermore, the overlapping resulting genes of the two integrated approaches were particularly enriched in biomass related pathways. A chromosome-wise over-representation analysis gave rise to the hypothesis that chromosomes number 2 and 4 probably carry a majority of the genes involved in biomass heterosis in the early development of Arabidopsis thaliana.

Figure 1. Setup of the over-representation analysis.

It is test if a systems biological approach towards heterosis and a quantitative genetics approach point to similar genomic regions influencing biomass heterosis in the early development of Arabidopsis thaliana.

Figure 2. ORA results for gene lists of two different approaches towards biomass heterosis.

: number genes in gene set. (A) For gene sets of genes, both hybrids show a significantly larger overlap between test set (3133 genes determined in QTL mapping experiments for biomass heterosis [5]) and gene set (determined in systems biological network analysis [4]) than expected for a random test set for MPH as well as BPH. (B) Fraction of how much more genes are observed in the overlap between test set and each of the gene sets than expected if this overlap would be a chance event relative to the expected overlap. The over-representation is significant (A) but not very strong (B).

Figure 3. ORA with original and several random gene sets.

This study was performed to confirm a significant over-representation between genes of two different approaches. (A) shows for C24×Col-0 MPH the number () of genes in the overlap between 3133 genes in the test set and genes in the “original” gene sets (black) and in 50 randomly chosen gene sets (gray), respectively. For the calculated empirical resampling p-values on the basis of 1000 randomly chosen gene sets confirm a significant over-representation of the gene set in the test set (B).

Figure 4. ORA for gene sets containing genes of only one of the five Arabidopsis thaliana chromosomes.

: number genes in gene set. (A)+(B): chromosome 1; (C)+(D): chromosome 2; (E)+(F): chromosome 3; (G)+(H): chromosome 4; for chromosome 5 no over-representation at all was observed and, therefore, no plots are shown. The probabilities of having as many or more genes in the overlap of random test sets to the gene set than observed for the experimental data are shown on the left side. The proportion of genes that were more in the overlap than expected by chance are shown on the right side. Genes on chromosomes 2 and 4 show a significant over-representation between the results of the two different approaches towards biomass heterosis in early development.

Figure 5. Relative frequencies of top ranked genes according to systems biological approach vs. Kosambi cM.

Relative frequencies (fraction between the number of detected 3000 top ranked genes of C24×Col-0 MPH of the systems biological analysis and all known genes in Arabidopsis thaliana at a specific section on each chromosome) are plotted against the genetic distance (Kosambi cM). (A–E): chromosome 1–5. The gray boxes are the QTL candidate genomic regions contributing to biomass heterosis . The horizontal lines are the median values of the relative frequencies per chromosome. For chromosomes 2 and 4 the relative frequency exceeds the median relative frequency especially in the area of the detected QTL. This is in line with the detected significant over-representation on these chromosomes of the genes detected in the systems biology approach in the genes from the QTL study.