Gene coexpression and functional enrichment analysis
The co-expression analysis for genes in embryo and endosperm was conducted using MeV (Howe et al., 2014). Notably, we used the Z-score value to calculate the relative expression levels before running the MeV for a tissue. These Z-score values were used as input for MeV and each individual dataset was clustered using the k-means method and Pearson correlation coefficients among the genes. Eventually, through Gene Ontology (GO) annotation (Ashburner et al., 2000), genes were annotated into the functional categories, and functional enrichment analysis was performed using the phyper function in R software with default setting (R Core Team, 2013). P-value was used as a filter mode, and significant GO categories (P-value <0.01) were identified. In addition, embryo or endosperm specific genes were identified by Z-score value. A gene was declared as embryo or endosperm specific if it had a Z-score above 2 in at least one of the embryo or endosperm samples as compared to all the same kind samples.