Figure 2

Length and GC3 content of ORFans and non-ORFans. While ORFans are statistically significantly shorter than non-ORFans in the majority of genomes, they have statistically significantly lower GC3 content than non-ORFans in a smaller fraction of the genomes. We applied the Wilcoxon nonparametric test to individual genomes. The null hypotheses were that ORFans are not shorter than non-ORFans, and that they do not have lower GC3 content than non-ORFans. For each genome, we compared the protein length and GC3 content distributions of its ORFans and non-ORFans. A p-value < 0.05 rejects the null hypothesis, and significantly supports the alternative hypothesis. A p-value > 0.95 supports the null hypothesis and a p-value between the two values means no significant difference. Only genomes with at least 30 ORFans and 30 non-ORFans were considered. The upper panels show the results of the length test on prokaryotes (left) and viruses (right). The lower panels show the results of the GC3 content test on prokaryotes (left) and viruses (right). Clearly, while the lower GC3 content of ORFans is statistically significant for about 70% of the bacterial genomes, it is statistically significant only for 27% of the viral genomes.