Figure 4.

Intepro domain sharing and expression correlation of isoforms. (A) Number of shared InterPro domains as a function of the number of shared GO terms. Brown (Isoform): boxplots for the number of shared InterPro domains between isoforms that share at least one domain (y-axis) as a function of different numbers of shared GO terms that are assigned by the isopret algorithm (x-axis), from 0 to 10 shared GO terms. The median of the number of shared InterPro domains increases with the number of shared GO terms. Blue Gene): the same boxplots for the number of shared InterPro domains, for different numbers of shared GO terms, where all the GO terms of a gene are assigned to all of its isoforms. While the isopret-assigned GO terms are selected from the gene’s GO terms and Interpro2GO, the correlation between the count of shared GO terms and the count of shared InterPro domains is higher for the former (Kendall’s tau 0.38 versus 0.18, respectively). (B) Correlation of expression level of transcripts as a function of the number of shared GO terms, for gene- and isoform-level assignments. Expression correlation as a function of the number of shared GO terms increases for both types of assignment, but for isoform-level it is consistently higher. Supplementary Fig. S6 shows a breakdown of the number of shared GO terms to the three subontologies MF, BP and CC