GeneRunner 5.0 was used to determine the size length of the coding sequence and the peptide chain.
Coding fRNAs tend to be evenly distributed within coding regions (the average relative position for a coding fRNA is 0.51±0.29 of the length of the coding region).
Random sequences containing 2.7×107 for E. coli and 2.5×107 codons for B. subtilis (20-times the length of the coding sequences in each genome) were generated.
Since 3-dimensional protein structure and binding sites determine protein functionality, the length of the coding sequence seems to be of primary importance, and therefore the variation in the transcript length may be constrained.
It has been shown that translation is more costly than transcription [20], and the shorter length of the coding sequences in housekeeping genes is likely the result of selection for economy of translation.
For our functional genomic analysis, we define a new variable, percent 'fRNA coverage,' which is the length of the coding region for a gene overlapping a predicted fRNA divided by the length of that gene.
The length of the coding region could play a role: longer coding regions correspond to longer protein chains and a longer translation time, thus leading to a greater chance of a proteolytic event.
For example, it is possible that the length of the coding and of the noncoding sequences are strongly related to each other because of structural constraints imposed by the process of splicing [18], such that the coding sequence length only has a relationship with expression because it is tightly tied to the gene noncoding sequence.
We therefore normalized the seed number in coding regions by dividing the total number of seeds in coding regions by the length of the coding sequence but still were not able to find any clear correlation between normalized seed numbers and regulation.
The results showed that the average number of exons per gene was directly proportional to the average length of the coding sequence per gene, both of which were directly proportional to the average length of the complete split-gene (Figure 4).
Expression studies done using uncorrected data from EST-libraries or indexes of codon bias [38], [39] are highly problematic and most likely biased when investigating intron density because both measures are strongly influenced by the length of the coding sequence [9], [28], [42].
