2. I feel somewhat inconvenient that the author includes intron number as one of the measures to gene compactness.
Partial correlation analyses suggest (table 3, Cases 3 5) that none of these correlations were driven by of the apparent pairwise covariation between these gene features with GC3 (which covariates with ω) and chromosome size (which covariates with measure of gene compactness).
We initially considered the following seven measures of gene compactness: the number of exons ('exon number'), lengths of coding exons and introns ('CDS length' and 'intron length'), lengths of untranslated regions (UTRs) ('5′-UTR' and '3′-UTR') and mean sizes of exons and introns ('exon size' and 'intron size').
> -wrap-foot> > -wrap-foot> The pairwise relationship between ω and several measures of gene compactness including the number of introns, total length of introns, and the length of the protein sequence are all significantly negative, in agreement with previous analysis of nonavian species (table 2, Larracuente et al. 2008; Choi and Hannenhalli 2013).
In mammals, gene compactness, measured by average length of introns or length of UTRs, has a stronger influence on protein evolutionary rates than does gene expression level, as mammalian proteins encoded by a more compact gene evolve more rapidly (Liao et al. 2006).
In yeasts, mRNA abundance is the predominant factor determining the rate of protein evolution (Drummond et al. 2006), whereas in mammals, gene compactness, measured by averaged length of introns or untranslated regions (UTRs), has a stronger influence on protein evolutionary rates compared with the abundance of mRNA (Liao et al. 2006, 2010).
We observe a generally non-linear relationship between gene compactness measures and gene expression levels (Fig. 1), as reported previously (20, 42).
> -wrap-foot> Recently, it has been shown that gene compactness measures have non-linear associations with expression levels (20), which is also the pattern we observe (Fig. 1).
Carmel and Koonin (20) examined the relationship between levels of gene expression and gene compactness measures and concluded that this relationship is 'universally non-monotonic', where genes with intermediate expression levels tended to be the least compact in four distantly related species (human, Drosophila melanogaster, Arabidopsis thaliana and Caenorhabditis elegans).
What causes such non-linear relationships between gene compactness variables and gene expression measures?
