ORMA: a tool for identification of species-specific variations in 16S rRNA gene and oligonucleotides design

Severgnini, Marco; Cremonesi, Paola; Consolandi, Clarissa; Caredda, Giada; De Bellis, Gianluca; Castiglioni, Bianca

The ability to identify 'fingerprint' positions within a set of homologous sequences, like those of 16S rRNA gene, is the main feature of ORMA. To achieve optimal results, the starting set of sequences should be carefully selected, because, if sequences are characterized by many lowsimilarity regions, the determination of terminally discriminating position could be biased by badly aligned subsequences. In that case, a different algorithm (actually not included, but under development) for the determination of detection probes by means of the hybridization strategy, can be more appropriated. On the other side, using sequences nearly identical one to each other can cause the opposite behavior, where no discriminating positions can be determined. A careful grouping of the sequences in clusters (as we did for both of our examples, building 18 consensus out of 352 sequences in cyanobacteria dataset and 13 consensus out of 752 sequences in milk pathogens dataset) is strongly suggested. In this latter application three rounds of design were applied, in order to compensate the non-perfect homogeneity of some species. Experimental results demonstrated the correctness of this approach and the specificity of the probe pairs obtained with this design strategy. Experimental data on the 16S rRNA cyanobacteria and milk-pathogens dataset demonstrated that ORMA specifically addressed discriminating positions within a set of highly similar sequences. Nonetheless, our tool identified a total of 192 and 392 candidate positions, respectively. The intra-and intergroup scores were demonstrated to be very helpful in determining the best probes for discrimination and avoiding cross-talk between species. ORMA is a bioinformatic tool for the search and determination of single-discriminating positions among a set of highly homologous sequences and represents a The scale varies between non-significance (>0.05) to high-significance (<0.005). The line 'Other' represents the mean of all the remaining Zip-codes in the universal arrays that were not associated to any actual probe. Complete association between samples numbers and names is given in Supplementary Table 2. significant improvement from other contexts where enzyme-based techniques have been employed on already known single-nucleotide polymorphisms (SNPs) (39) or on entire subsequences (11) . This unique feature makes ORMA completely different from all the other available software for probe design in detection experiments. During the past years, academic software for species detection have been developed. ProDesign (13) is a tool based on a 'spaced seed algorithm' for the determination of probes capable of discriminating multiple pathogenic species, at different hierarchical levels. Similarly, YODA (14) performs design tasks on complete genomes against non-target species. TOFI-beta (15) implements a suffixtree-based algorithm for isolating suitable candidate probes from a target genome and filters the list according to thermodynamical and specificity requirements. These three software are implemented for the design of probes for hybridization-based detection assays. PathogenMIPer (16) , instead, is based on a different strategy (i.e. molecular inversion assays), which starts from the selection of unique sequences on a reduced dataset and then does a global comparison to all those potentially matching.