Figure 1

Figure 1. Logo analysis on spoligotypes associated with Mycobacterium bovis. The Wadsworth Center database contains 28 unique spoligotypes from strains of M. bovis. Panel A illustrates the raw hybridization data followed by the...

Figure 1 illustrates logo analysis with spoligotyping data. Spoligotyping identified 28 different spoligotypes associated with M. bovis from 43 isolates in our collection. The spoligotypes were then coded for sequence logo analysis (Figure 1A). Letter designations were chosen for compatibility with the WebLogo program as described in Methods. The tallest x and o characters represent areas of absolute concordance between the patterns chosen for analysis. Where differences occur, the ratio of those spoligotypes showing hybridization to those that do not is represented by relative height differences of the characters in that column. The resulting spoligologo (Figure 1B) shows that spacers 20, 25, 26, and 38 are present in all 28 spoligotype patterns. The absence of spacers 39–43 in these spoligotype patterns is consistent with an identification of M. bovis (13). The greatest polymorphism between the patterns appears in spacers 3 through 16 (Figure 1B). Spoligotype patterns probably evolve through the loss of spacer sequences through a variety of mechanisms (7,20). We try to extrapolate back to the hypothetical founder of these M. bovis isolates, which we believe had spacers 1–38 and not 39–43. No isolate with this probable M. bovis founder spoligotype has yet been observed in the Wadsworth Center spoligotype database.

Figure 2

Figure 2. Logo analysis on nine different spoligotypes observed for Mycobacterium tuberculosis isolates from Vietnam-born patients in Massachusetts demonstrating fewer than six copies of IS6110 by RFLP analysis. Legend: x= hybridization observed to...

Foreign-born persons, especially those from regions with high TB case rates, are of concern for TB transmission in the United States (21,22). The low-band M. tuberculosis strains from Vietnam-born patients, which we selected, produced another spoligologo pattern (Figure 2). The ability of spoligologo analysis to collapse even a small selection of spoligotype patterns from a select group of strains (Figure 2A) into the possible founder spoligotype can be observed in the summary (Figure 2C). However, additional typing methods would be required to verify that the strains are related, rather than exhibiting convergent evolution of their respective spoligotypes.

As previously suggested (5,7,23), M. tuberculosis strains that generally contain spacers 33–43 may form a family that is an intermediate lineage between the Beijing (11,12) and non-Beijing M. tuberculosis strain families, such as Haarlem (2).

Spoligotyping, microarrays, and DNA-chips are all examples of reverse-hybridization array-based assays. Although microarrays and DNA-chips can contain thousands of bits of data, the principle behind them is similar to that of the spoligotype assay, which uses a simple 1 x 43 array. Array-based assays use reverse hybridization in which a labeled sample is probed against a series of proteins or nucleic acids that are bound to a solid support, such as a nylon membrane or silica. The result for each potential binding event can be recorded as yes or no. The binary nature of array-based assays allows the data to be analyzed usefully with algorithms associated with motif recognition, such as sequence logo analysis. The relative low cost and simplicity of the spoligotype assay means it can be performed by many laboratories and the digital nature of the data facilitates the exchange of information among researchers.

The growth in the availability of array-based assays has outpaced the ability of conventional software analysis packages to provide every possible method of analysis. Customized versions of software are extremely expensive, and researchers, who want to implement these protocols without specialized software, lack methods of collating the large amounts of data. Problems arise when attempts are made to judge the significance of similar but nonidentical array data. Identifying possible families in these array patterns may be important in understanding the evolution and spread of pathogens such as M. tuberculosis. Our method for collapsing array-based data can be used to find and present patterns or signature motifs in these types of data.

As previously noted (1), cluster analysis of large RFLP databases is difficult for a combination of reasons, including software failure and intralaboratory variations. Digital data, such as spoligotyping, mycobacterial interspersed repetitive unit (24), and variable number tandem repeat analyses (25), will probably form the basis for any large DNA fingerprinting projects in the future (26).

The design of large genotyping projects should include multiple methods (2,3,10,23). Strains that cluster by one typing method must be analyzed by other methods to ensure that the groupings represent clusters of true relatedness and not cases of convergent evolution.

Bioinformatic analyses, like logo analysis, may prove useful in obtaining further data from the large M. tuberculosis complex DNA typing databases already in existence. Spoligologo analysis is a graphic method of presenting similar spoligotypes that may elicit useful insights into the geographic spread of tuberculosis. Potential families of TB strains could be identified on the basis of their logo; these strains could then be analyzed by additional DNA-typing methods to confirm the relationship, followed examining relevant patient data (e.g., country of birth). A similar analysis was performed for the M. tuberculosis W-Beijing family (11) that helped elucidate the evolution of a multidrug–resistant strain in New York City. Spoligologo analysis could help identify more of these families, determine their global origin, and evaluate their spread.

Determining the sources and spread of tuberculosis is an important tool in preventing further infections. Understanding the geographic origin of an M. tuberculosis DNA fingerprint could be useful, especially in understanding the sources and spread of strains in the U.S. foreign-born population, among whom differentiating recently transmitted disease from reactivation of a past exposure can be difficult.