Stability distribution in the phage lambda-DNA double helix: a correlation between physical and genetic structure

DNA-energetics-based analyses suggest additional genes in prokaryotes

We present here a novel methodology for predicting new genes in prokaryotic genomes on the basis of inherent energetics of DNA. Regions of higher thermodynamic stability were identified, which were filtered based on already known annotations to yield a set of potentially new genes. These were then processed for their compatibility with the stereo-chemical properties of proteins and tripeptide frequencies of proteins in Swissprot data, which results in a reliable set of new genes in a genome. Quite surprisingly, the methodology identifies new genes even in well-annotated genomes. Also, the methodology can handle genomes of any GC-content, size and number of annotated genes.

A stitch in time: efficient computation of genomic DNA melting bubbles

Background

It is of biological interest to make genome-wide predictions of the locations of DNA melting bubbles using statistical mechanics models. Computationally, this poses the challenge that a generic search through all combinations of bubble starts and ends is quadratic.

Results

An efficient algorithm is described, which shows that the time complexity of the task is O(NlogN) rather than quadratic. The algorithm exploits that bubble lengths may be limited, but without a prior assumption of a maximal bubble length. No approximations, such as windowing, have been introduced to reduce the time complexity. More than just finding the bubbles, the algorithm produces a stitch profile, which is a probabilistic graphical model of bubbles and helical regions. The algorithm applies a probability peak finding method based on a hierarchical analysis of the energy barriers in the Poland-Scheraga model.

Conclusion

Exact and fast computation of genomic stitch profiles is thus feasible. Sequences of several megabases have been computed, only limited by computer memory. Possible applications are the genome-wide comparisons of bubbles with promotors, TSS, viral integration sites, and other melting-related regions.

Partly melted DNA conformations obtained with a probability peak finding method

Peaks in the probabilities of loops or bubbles, helical segments, and unzipping ends in melting DNA are found in this article using a peak finding method that maps the hierarchical structure of certain energy landscapes. The peaks indicate the alternative conformations that coexist in equilibrium and the range of their fluctuations. This yields a representation of the conformational ensemble at a given temperature, which is illustrated in a single diagram called a stitch profile. This article describes the methodology and discusses stitch profiles vs the ordinary probability profiles using the phage lambda genome as an example.

GC/AT-content spikes as genomic punctuation marks

Large-scale analysis of the GC-content distribution at the gene level reveals both common features and basic differences in genomes of different groups of species. Sharp changes in GC content are detected at the transcription boundaries for all species analyzed, including human, mouse, rat, chicken, fruit fly, and worm. However, two substantially distinct groups of GC-content profiles can be recognized: warm-blooded vertebrates including human, mouse, rat, and chicken, and invertebrates including fruit fly and worm. In vertebrates, sharp positive and negative spikes of GC content are observed at the transcription start and stop sites, respectively, and there is also a progressive decrease in GC content from the 5' untranslated region to the 3' untranslated region along the gene. In invertebrates, the positive and negative GC-content spikes at the transcription start and stop sites are preceded by spikes of opposite value, and the highest GC content is found in the coding regions of the genes. Cross-correlation analysis indicates high frequencies of GC-content spikes at transcription start and stop sites. The strong conservation of this genomic feature seen in comparisons of the human/mouse and human/rat orthologs, and the clustering of genes with GC-content spikes on chromosomes imply a biological function. The GC-content spikes at transcription boundaries may reflect a general principle of genomic punctuation. Our analysis also provides means for identifying these GC-content spikes in individual genomic sequences.

GeneFizz: A web tool to compare genetic (coding/non-coding) and physical (helix/coil) segmentations of DNA sequences. Gene discovery and evolutionary perspectives

The GeneFizz (http://pbga.pasteur.fr/GeneFizz) web tool permits the direct comparison between two types of segmentations for DNA sequences (possibly annotated): the coding/non-coding segmentation associated with genomic annotations (simple genes or exons in split genes) and the physics-based structural segmentation between helix and coil domains (as provided by the classical helix-coil model). There appears to be a varying degree of coincidence for different genomes between the two types of segmentations, from almost perfect to non-relevant. Following these two extremes, GeneFizz can be used for two purposes: ab initio physics-based identification of new genes (as recently shown for Plasmodium falciparum) or the exploration of possible evolutionary signals revealed by the discrepancies observed between the two types of information.

Variations of the mononucleotide and short oligonucleotide distributions in the genomes of various organisms

We calculated the variation coefficients of the mononucleotide and short oligonucleotide distributions in over 1700 long genomic sequences originating from six organisms to demonstrate that the human and Escherichia coli genomic sequences were the least and the most uniform, respectively. The most non-random genomic distributions were exhibited by the four canonical nucleotides, followed by the strong and weak nucleotides, while the distributions of purine or pyrimidine nucleotides and especially the distributions of (A+C) and (G+T) were significantly more uniform even in the human genome. In the human and mouse genomes, the highest coefficients of variation were further observed with the oligonucleotides where CG was combined with the strong nucleotides while its combination with the weak nucleotides significantly decreased the variation which, however, was still very high. High variation was also exhibited by the remaining oligonucleotides composed exclusively of the strong nucleotides or those containing only weak nucleotides. On the other hand, the distributions of oligonucleotides containing similar and especially the same numbers of the strong and weak nucleotides, but no CG or TA dinucleotide, were the most uniform. The information following from the present analysis will be useful not only in the identification of important genomic regions but also in computer simulations of the genomic nucleotide sequences in order to trace and reproduce the pathways of genome evolution.

Characteristic features of thermal stability map of DNA in Escherichia coli and eukaryotic genes

Distribution of double-helix thermal stability of Escherichia coli and eukaryotic DNAs was analyzed. The results confirmed the previous propositions based on the study of the stability distribution in phage DNAs: (1) stability fluctuation appears near the boundaries of protein coding regions (PCRs) and non protein coding regions (NPCRs); (2) PCRs have less fluctuation than NPCRs. The present analysis also revealed that the local G + C content is lower in the beginning of PCRs of E. coli than the average G + C content of PCR and that deviations in the amino acid composition and the third letter usage PCRs are involved in the low G + C content; the biological meaning of this is discussed in relation to mRNA structure.

Stability of RNA stem-loop structure and distribution of non-random structure in the human immunodeficiency virus (HIV-I)

The stability of potential RNA stem-loop structures in human immunodeficiency virus isolates, HTLV-III and ARV, has been calculated, and the relevance to the local significant secondary structures in the sequence has been tested statistically using a Monte Carlo simulation method. Potentially significant structures exist in the 5'non-coding region, the boundary regions between the protein coding frames, and the 3' non-coding region. The locally optimal secondary structure occurring in the 5' terminal region has been assessed using different overlapping segment sizes and the Monte Carlo method. The results show that the most favorable structure for the 5' mRNA leader sequence of HIV has two stem-loops folded at nucleotides 5-104 in the R region (stem-loop I, 5-54 and stem-loop II, 58-104). A large fluctuation of segment score of the local optimal secondary structure also occurs in the boundary between the exterior glycosylated protein or outer membrane protein and transmembrane protein coding region. This finding is surprising since no RNA signals or RNA processing are expected to occur at this site. In addition, regions of the genome predicted to have significantly more open structure at the RNA level correlate closely with hypervariable sites found in these viral genomes. The possible importance of local secondary structure to the biological function of the human immunodeficiency virus genome is discussed.

Vestiges of lost introns in the thermal stability map of DNA

The absence of introns in prokaryotic genes has been explained by intron loss on various bases. Here we report another piece of evidence on intron loss, which was found in the thermal stability map of DNA. We calculated the local melting temperature of cDNA sequences and found that (i) gaps in thermal stability tend to occur near intron positions with a statistical significance, and (ii) one-third of the gaps far from intron positions can be assigned to lost introns. From these results we conclude that the gaps of thermal stability in protein coding regions are the vestiges of lost introns.

Distribution and evolution of sequence characteristics in the E. coli genome

The mean (G + C) composition (51.0%) and standard deviation (+/- 3.8%) of published DNA sequences accounting for 10% of the E. coli genome is in excellent agreement with the principal overall distribution determined by high resolution melting. While differences in base and neighbor characteristics are small and uniform throughout all regions of the genome, it is found that the (G + C) content of sequences varies in segmented fashion within boundaries corresponding to coding (53% G + C) and noncoding (46% G + C) regions; with variances in the latter being six-fold greater than in coding regions. The variance in different regions shows a strong negative dependence on (G + C) content of the region, reflecting the condition that A-T and G-C base pairs are preferred neighbors of A-T and C-G pairs, respectively; with the bias increasing with decreasing (G + C) content. Neighbor analysis indicates the most extreme positive biases occur in AA, TT, GC and CG throughout all regions, but particularly in noncoding regions. Extraordinary numbers of oligomeric strings of (A)n, etc., are the further consequence of this bias. These and other characteristics point to the existence of inherent biases in neighbor frequencies levied during replication or repair, and which reflect, in turn, neighbor influences during mutation. The bias in codon usage noted by Grantham and others is seen here as due, in part, to the adaptation of coding sequences to this microenvironment through selection among synonymous codons so as to preserve inherent neighbor biases.