3. Results and Discussion

3.1. Information Obtained for Interpreting Virtual Hybridization Results
Information produced by the Virtual Hybridization software is stored in an output file in text format; this includes: file name, total number of genomes hybridized, number of experimental probes, name of the file where the probes set is stored, name of the genome hybridized, number of times that the same probe hybridized across the genome, probe number, probe sequence, position in the genome where the probe hybridized, complementary sequence, and Δ°G between the probe and the target sequence. Comparison of the virtual hybridization patterns obtained with the direct method yielded an average of 1528 potential hybridization sites, and 4603 for the extended method (Table 2).
microarrays-04-00084-t002_Table 2 Table 2  Number of signals obtained with direct or extended methods for Bacillus anthracis.   * Upon virtual hybridization with 15,264 probes, using a Δ G° value between −19.53 and −11.67 (kcal/mol). ** Upon virtual hybridization with 15,264 probes, using a Δ G° value between −19.53 and −13.01 (kcal/mol).

3.2. Bacillus Anthracis Virtual Genomic Fingerprints
Results from the virtual hybridization of the bacterial genomes with the UFC-13 include: the probes with which hybridization occurred, the sites in the genome where hybridization took place, stability values for the heteroduplexes formed, and the sequences involved. From these data, an image of the virtual hybridization pattern of each organism is generated; showing the sites on the microarray of UFC-13 probes where binding occurred and those sites where no hybridization occurred. The overall image of each organism’s virtual hybridization pattern in the DNA sensor constitutes its genomic fingerprint. Microarray_pic shows a genomic fingerprint of each Bacillus anthracis strain. This image represents an in silico DNA microarray for a given organism, together with the specific probes used in hybridization experiments. This tool shows the set of 15,264 probes on a microarray as spots, color‑coded to identify those probes that hybridized (Figure 3).
Figure 3  Fingerprints of two bacteria (A) Bacillus anthracis Ames Ancestor (35.26 GC%, 5.5 Mb); (B) Bacillus anthracis Kruger (35.1 GC%, 5.47 Mb); (C) Combination and overlap fingerprint of both microorganisms.

3.3. Bacillus Anthracis Analysis
Data on potential sites of virtual hybridization were used to construct the genomic fingerprints distance matrix. Two analyzes were conducted, using the direct and the extended method; similarity indices and distances for the eight strains studied were calculated. Phylogenomic trees were obtained for both, the direct and the extended method. In the resulting trees, the bacterial groups are arranged according to their similarities and differences.

3.4. Distance Table
The fingerprints were compared to calculate similarity measures between them. In order to accurately identify a given organism from fingerprint results, each organism should yield a specific virtual hybridization pattern and the between-patterns similarity should be related to their genomes (as to length and G + C content). The number of signals shared by fingerprint patterns can be used to estimate similarity indices and distances between genome sequences.
Table 3 shows the distance values between the eight Bacillus anthracis strains studied, out of an average of 4603 potential sites. When a given organism is compared to itself, the distance value is zero but when it is compared to another organism, the distance value increases (0 < score ≥ 1), in relation to the genomic difference between the strains. The minimum value (0.000017) found in this study (shown in yellow in Table 3) corresponds to the comparison between the genomic fingerprints of Bacillus anthracis str. Ames and Bacillus anthracis str. Ames Ancestor (allowing only one mismatch and using a ΔG° between −19.53 and −11.67 kcal/mol). The largest distance value (0.001088) was obtained for Bacillus anthracis str. Kruger.
microarrays-04-00084-t003_Table 3 Table 3  Distances calculated from the extended method results.   When thermodynamic conditions were made stricter by using a ΔG° between −19.53 and −13.01 kcal/mol for the eight genomes, with an average of 1528 signals, the UFC was not sensitive enough to discriminate between highly similar strains. A minimum distance value of zero (green) was obtained for the comparison between the Bacillus anthracis str. Ames and Bacillus anthracis str. Ames Ancestor strains. This means that the method cannot discriminate these two strains. The largest distance value (0.001261) was obtained for the comparison between these two strains and Bacillus anthracis str. Kruger; the latter is the strain that is most different to the others (Table 4).
microarrays-04-00084-t004_Table 4 Table 4  Distances calculated from the direct method results.

3.5. Bacillus Anthracis UFC-13 Trees
A taxonomic tree is a visual representation of the degree of relatedness that the Bacillus anthracis strains hold either by descent from a common ancestor or by high similarity. Distances between all possible pairs of Bacillus anthracis fingerprints were calculated based on the number of signals where virtual hybridization between the probes and the genome took place. Each genome fingerprint has its own virtual hybridization signals and these are compared with other fingerprints. Then, a distance matrix comparing those signals is calculated. The tree is finally constructed from the distance values calculated from the number of signals of each fingerprint, using the Neighbor-Joining method in the PHYLIP 3.61 software package (Figure 4).
Figure 4  Phylogenomic trees obtained by comparing genomic fingerprints with UFC-13. The similarity between the two methods used for classifying the Bacillus anthracis strains shows the degree of sensitivity necessary to distinguish between closely similar strains. (A) Virtual Hybridization of 15,264 probes under relaxed conditions, with ΔG° values between −19.53 and −11.67 kcal/mol; (B) Virtual Hybridization of 15,264 probes under strict conditions, with ΔG° values between −19.53 and −13.01 kcal/mol. Both taxonomic trees show that Bacillus anthracis str. Kruger B is the one most distant from Bacillus anthracis str. Ames.   Genome comparison is important for understanding the biological functions determined by genomic information. Genome comparison methods such as FASTA and BLAST have helped to understand the function of thousands of sequences and the evolutionary relationships among different bacteria.The genomes of Bacillus anthracis str. Ames and Bacillus anthracis str. Ames Ancestor were aligned using the MUMmer 3.0 software; the number of substitutions, insertions and deletions were determined, as well as the number of nucleotides involved in each of those, in order to quantify the differences between the two genomes. The results revealed only 134 differences out of 5,227,293 bases for a 99.9975% similarity between these two genomes [17,18] (Figure 5).
Figure 5  Dot plot showing the alignment of 5,227,293 bp of Bacillus anthracis str. Ames ancestor (x-axis) and 5,227,419 bp of Bacillus anthracis str. Ames (y-axis). Alignment segments are represented as lines or dots. The blue diagonal line denotes the similarity between the sequence and the genomic location of the two strains. Scattered points represent coinciding sequences located on different sites in the genomes.

4.