A statistical analysis of the three-fold evolution of genomic compression through frame overlaps in prokaryotes

Background: Among microbial genomes, genetic information is frequently compressed, exploitingredundancies in the genetic code in order to store information in overlapping genes. We investigatethe length, phase and orientation properties of overlap in 58 prokaryotic species evaluating neutraland selective mechanisms of evolution.Results: Using a variety of statistical null models we find patterns of compressive coding that cannot be explained purely in terms of the selective processes favoring genome minimization ortranslational coupling. The distribution of overlap lengths follows a fat-tailed distribution, in whicha significant proportion of overlaps are in excess of 100 base pairs in length. The phase of overlap– pairing of codon positions in complementary reading frames – is strongly predicted by thetranslation orientation of each gene. We find that as overlapping genes become longer, they havea tendency to alternate among alternative overlap phases. Some phases seem to reflect codonpairings reducing the probability of non-synonymous substitution. We analyze the lineagedependentfeatures of overlapping genes by tracing a number of different continuous charactersthrough the prokaryotic phylogeny using squared-change parsimony and observe both cladespecificand species-specific patterns.Conclusion: Overlapping reading frames preserve in their structure, features relating tomutational origination of new genes, but have undergone modification for both immediate benefitsand for variational buffering and amplification. Genomes come under a variety of differentmutational and selectional pressures, and the structure of redundancies in overlapping genes canbe used to detect these pressures. No single mechanism is able to account for all the variabilityobserved among the set of prokaryotic overlapping genes but a three-fold analysis of evolutionaryevents provides a more integrative framework