Figure 3: Validation of AraNet by independent data sets.

(a) The predictive power of AraNet for 86 GO cellular component terms. (b) AraNet predictive power for 82 KEGG metabolic pathways (excluding isozymes). Bar-and-whiskers plots are as in Figure 2b–d. (c) Co-occurrence of mRNA transcripts across 20 root cell-types³¹ for network-linked genes. Genes linked in AraNet (AraNet, lane 6) were more co-expressed in each root cell-type than gene pairs from random networks (repeating the calculation for 100 randomized networks and plotting the distribution of the 100 resulting odds ratios (randomized, lane 1; see Supplementary Methods for details). This trend cannot be explained simply by the incorporation of Arabidopsis mRNA expression data into AraNet, as a version of the network lacking this data shows similarly high cell-type specificity (AraNet no AT-CX, lane 7). Compared to other previous Arabidopsis networks (lanes 2–5, Supplementary Table 3), AraNet shows 1.6- to 4-fold higher cell-type specificity. Subset of AraNet composed of only known-unknown genes shows high cell type–specificity as well, supporting predictability of AraNet using cell-specific expression data (lane 8). (d) AraNet shows predictability for genes affecting embryonic lethality or changes in seed pigmentation, as identified in the SeedGenes database³². AUC values are indicated in parentheses.