DE102005002672A1 - Differentiation of microorganisms in a sample, comprises extracting single cultures of microorganisms, receiving mass spectra on the cultures, and analyzing and comparing spectral details with reference spectra by mathematical similarity - Google Patents

Abstract

Differentiation of microorganisms in a sample comprises that single cultures are extracted from the sample by cultivation on selective species medium, and the method further comprises cluster analysis of the extracted culture from the sample by matrix assisted lasers desorption/ionization time of flight mass spectra (MALDI-TOF MS) with a range of 2000-20000 Dalton. Differentiation of microorganisms in a sample comprises that single cultures are extracted from the sample by cultivation on selective species medium, and the method further comprises cluster analysis of the extracted culture from the sample by matrix assisted lasers desorption/ionization time of flight mass spectra (MALDI-TOF MS) with a range of 2000-20000 Dalton. The culture and/or extracted culture are identical to the reference subspecies when the distance of the spectra is smaller than sigma 1. The culture has a new subspecies, when the distance from all reference spectra and the other extracted cultures from the sample is greater than sigma 2. The culture and/or extracted culture are closely related to a reference subspecies when the distance of the spectra is greater than sigma 1 and smaller than sigma 2. The mass spectra of the single culture are compared by a euclidean distance matrix with the reference spectra, where the spectrum is constructed from the overlay of six complete single spectra. The values sigma 1 and sigma 2 are extracted from the euclidean distance matrix of the reference spectra.

Description

The The invention relates to a method for the rapid differentiation of Microorganisms at subspecies level using MALDI (matrix assisted laser desorption ionization) -TOF (time of flight) mass spectrometry.

The Invention is applicable where rapid identification and Differentiation of microorganisms is required, for. B. in the Microbiology, human and dental medicine, veterinary medicine or biotechnology.

method to identify microorganisms using Matrix Assisted Laser Desorption / Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS) are published. These are based either on targeted detection one for the respective microorganism characteristic, known substance on their Mass peak in the spectrum (Allmaier et al., 1995) or on the comparative Analysis of reduced (Kallow et al., 2002) or entire mass spectra (Bright et al., 2002)

The previously known methods for the characterization of microorganisms using MALDI-TOF MS work up to the species level because the mass spectra of different Species are usually distinct and different from each other the occurrence of relatively fewer, typical peaks can be characterized.

A Differentiation at subspecies level over known, characteristic Substances over their mass peak in the spectrum implies that for each subspecies at least such a substance is known. Such substances are however only for very few Microorganism subspecies known (Hathout et al., 1999; Elhahnany et al., 2001; Hatout et al., 2003; McNall et al., 2003; Whiteaker et al., 2004). Such methods are thus not generally applicable.

The comparative analysis of the total mass spectra of various Microorganisms is very complex because the spectra have a lot of information contained, which must be extracted by suitable methods. Which Peaks or groups of peaks for which taxonomic level are characteristic can be do not recognize or predict a priori. At the subspecies level Although distinguishable spectra for various known bacterial subspecies observed and these could be assigned to the respective subspecies (Bright et al., 2002), but a systematic differentiation unknown cultures at subspecies level has not previously been described and is not possible with the methods described so far.

task The invention is to provide a method which is fast Differentiation of microorganisms at subspecie level allows. A species includes all organisms that belong to the same genus and bear species names. Different species typically differ in terms of its 16S rDNA by more than 3%. Conversely, subspecies (or tribes) Organisms that carry the same genus and species name and themselves in their 16S rDNA only very slightly, i. less than 3%, different.

• a. aus der Probe durch Kultivierung auf speziesselektivem Medium Einzelkulturen von Mikroorganismen gewonnen werden, die zur gleichen Spezies gehören,
• b. MALDI-MS-Spektren der aus der Probe gewonnenen Einzelkulturen aufgenommen werden,
• c. durch eine mathematische Ähnlichkeitsanalyse aus MALDI-MS-Referenzspektren von verschiedenen Mikroorganismensubspezies, die zur gleichen Spezies wie die aus der Probe gewonnen Einzelkulturen gehören, die Werte σ₁ (Homogenitätsgrenze) und σ₂ (Heterogenitätsgrenze) ermittelt werden,
• d. die MALDI-MS-Spektren der aus der Probe gewonnenen Einzelkulturen paarweise miteinander und mit den Referenzspektren durch eine mathematische Ähnlichkeitsanalyse verglichen werden.

According to the invention, this object is achieved by a method for differentiating microorganisms in a sample at the subspecies level by means of matrix assisted laser desorption / ionization time-of-flight mass spectrometry (MALDI-TOF MS),
in which

• a. obtained from the sample by culturing on species-selective medium individual cultures of microorganisms belonging to the same species,
• b. MALDI-MS spectra of the individual cultures obtained from the sample are recorded,
• c. by means of a mathematical similarity analysis from MALDI-MS reference spectra of different microorganism subspecies belonging to the same species as the individual cultures obtained from the sample, the values σ ₁ (homogeneity limit) and σ ₂ (heterogeneity limit) are determined,
• d. the MALDI-MS spectra of the individual cultures obtained from the sample are compared in pairs with each other and with the reference spectra by a mathematical similarity analysis.

If the distance of the spectra is smaller than σ ₁ , the single culture is identical to a reference subspecies or another single culture obtained from the sample.

If the distance to all reference spectra and to all spectra of the other individual cultures obtained from the sample is greater than σ ₂ , the single culture is a new subspecies.

If the distance of the spectra is greater than σ ₁ and less than σ ₂ , the single culture is closely related to a reference subspecies or other single culture derived from the sample.

The homogeneity limit σ ₁ corresponds to the maximum distance of different spectra of the same reference subspecies in multiple measurements of the reference subspecies.

The heterogeneity limit σ ₂ corresponds to the smallest observed distance between spectra of two different reference subspecies.

The inventive method characterized by the combination of cultivation of those contained in the sample Microorganisms on a species-selective medium, MALDI-TOF MS analysis the obtained individual cultures as well as differentiation of the usually very similar MALDI TOF-MS spectra using multivariate statistical analysis using different clustering methods.

By appropriate cultivation methods is first achieved that all off of a biological sample isolated individual cultures at the same time Species or species group. By combining MALDI-TOF MS and suitable procedures the mathematical spectral comparison is then achieved that the mass spectra of these individual cultures, despite their high similarities reproducible and differentiated with specifiable selectivity on subspecies level can be.

at the biological sample is, for example, a human one Saliva sample, a plaque sample or a smear through a swab the oral cavity won sample.

The microorganisms contained in the sample are first cultured. The cultivation is carried out on a selective medium that growth of a particular microorganism species or a species group and growth as possible many other microorganisms inhibits. Such a species-selective Medium is for example Mitis Salivarius Agar with bacitrazine. This culture medium is a selective medium for human pathogens Mutans streptococci.

Around to ensure, that, for example, only microorganisms are examined by mass spectroscopy, which belong to the species Streptococcus mutans, which are by selective Culture conditions obtained individual cultures preferably additionally by means of biochemical tests (e.g., the so-called "Colorful Series", a series of liquid cultures for determining physiological properties, e.g. the formation of acids from different sugars, with a positive test failure a pH shift and the color change of an added indicator causes) and / or over their Membranfettsäurespektrum (Kato, 1989; Stößer et al., 2000) at species level.

To accommodate the MALDI-TOF MS spectra, suitably 10 ⁵ to 10 ⁶ cells of a single culture are used. If there are not enough cells per individual culture, a further cultivation step in a selective or non-selective medium is necessary.

The Cells are collected by centrifugation. The cells will be if necessary, for example, cleaned with water and either directly on a MALDI-MS sample carrier applied and mixed with a suitable matrix solution or they are first with a matrix solution and then applied to a suitable MALDI-MS sample carrier.

When MALDI-matrix solution For example, α-cyano-4-hydroxy-cinnamic acid (HCCA) in acetonitrile / 0.1% trifluoroacetic acid (TFA) (2: 1). Preferably, the same single culture is repeated several times Positions on the sample carrier applied.

The MALDI sample carrier is inserted into a suitable MALDI-TOF mass spectrometer (eg Bruker Biflex III, N ₂ laser (wavelength 337 nm, pulse 5 ns, flight distance 1200 mm).) The measurements are suitably performed in linear positive mode (delay 400ns, voltage 20 kV) The spectra are preferably calibrated internally, eg the Peptide Mix (Sigma, Taufkirchen) is used for the calibration.

The MALDI-TOF MS spectra are preferably in a mass range of 2000 to 20,000 daltons recorded.

From each individual culture on the sample carrier are preferably several Single spectra recorded from which by overlay a MALDI-TOF-MS sum spectrum is formed. It is particularly advantageous to have at least six MALDI-TOF-MS sum spectra, the come from the same single culture, but in different positions on the sample carrier were superimposed to superimpose to a super spectrum. In this way is for Each single colony received a superspektrum with higher reproducibility.

Under the same conditions are also used for MALDI-TOF-MS super spectra of several known subspecies of the microorganisms in question added as a reference. The reference microorganisms belong to the same species as isolated from the biological sample individual cultures. Differentiation of subspecies of S. mutans in a biological Sample, for example, the subspecies of S. mutans like Ingbritt, JB 1600, NCTC 10449, OMZ 125, GS 5, SE 11, OMZ 175, LM 7 or QP 50-1 for reference.

Preferably be for each reference subspecies more, more preferably at least two independent Superspektren determined.

Peak lists are created from the super spectra obtained. The peak lists are transformed in a suitable way, eg log-transformed. Peaks of equal mass / charge within the different super spectra are determined by preferentially passing a mass window along the x-axis (m / z) of the superspec spectra. The center of this mass window represents the mean position of the comparable peaks in all spectra studied. The window width is determined by the precision of the experimental m / z determination. A linear relationship to m / z is assumed. Thus results: Window width = absolute width + relative width · peak mass. (Formula 1)

The Mass window is assigned to each mass peak. Windows in different superspectra containing the same peaks summarized. These peaks are indicated by their mean m / z and their medium intensity represents. overlapping Windows are resolved with half the window width.

By the method described, for example, a matrix η _{ij is} constructed. Their lines i are determined by the middle peak masses, their columns j by the respective superspec spectrum. In the absence of a peak in a superspec spectrum, η _ij = 0 is assumed.

The similarity the superspectral spectrum is preferably given by a Euclidean distance matrix shown.

η_kj stellt die Intensität des idealisierten Peak k im Spektrum j dar, w_kmn ist ein Maß dafür, ob die k-te Peak Masse in einem Spektrum einen vorgegebenen Grenzwert überschreitet. Unter Berücksichtigung einer variablen Anzahl Peaks je Superspektrum wird der Distanzkoeffizient d_mn korrigiert:

wobei n_m und n_n die Anzahl der Peaks in den normalisierten Superspektren m und n darstellen, die einen vorgegebenen Grenzwert überschreiten. Diese Korrektur ist entscheidend, um den Einfluss der unterschiedlichen Anzahl Peaks je Superspektrum auszugleichen.The Euclidean distance coefficient d _mn compares two super spectra m and n, and is defined as:

η _kj represents the intensity of the idealized peak k in the spectrum j, w _kmn is a measure of whether the k-th peak mass in a spectrum exceeds a predetermined limit. Taking into account a variable number of peaks per super spectrum, the distance coefficient d _{mn is} corrected:

where n _m and n _{n represent} the number of peaks in the normalized super spectra m and n that exceed a predetermined threshold. This correction is crucial to compensate for the influence of the different number of peaks per superspace.

Two distance measures, σ ₁ and σ _{2, are} determined from the distance matrix of the multiply determined reference super spectra of the subspecies of a species: σ ₁ corresponds to the maximum distance between multiple measurements of the reference subspecies and is a measure of the minimum resolvable distance of two subspecies (Homogeneity limit). This means that isolates whose super-spectra have a distance smaller than σ ₁ to one another can not be differentiated and must therefore be regarded as identical.

σ ₂ corresponds to the smallest observed distance between superspec- tive spectra of two differently known subspecies of a species (heterogeneity limit). Then, σ ₂ and σ ₁ are compared with each other. The method is only valid if σ _{2 is} greater than or equal to σ ₁ .

One Comparison of the distances of the super spectra of the individual cultures with the superspectral spectra of reference subspecies to other species belong, allows a favorable review of the correct assignment of the individual culture to the species to be examined.

to Subspecies population analysis of biological samples from which single cultures of a species are obtained, the superspectrics of the individual cultures first a pairwise comparison with each other and with the super spectra the already known reference subspecies subjected. Following is preferred a global similarity analysis carried out.

For each individual culture, it is checked whether their superspec spectrum has a distance smaller than σ ₁ to that of one of the reference subspecies. If the distance is smaller than σ ₁ , the individual culture is identical to the respective reference subspecies.

If this is not true, it is checked whether the superspectra of each of two individual cultures to each other have a distance smaller σ ₁ . Then they are identical to each other and will be replaced by their centroid in further analysis. The centroid is the midpoint of the multidimensional space defined by the mass peaks of the two spectra. The centroid is an idealized spectrum that contains all the masses that occur in both spectra. Each of these masses is assigned an intensity that is calculated as the average of the intensities for that mass.

Let S ₁ S ₂ be 2 spectra forming a cluster.

Let m ₁ ..... m _n be the masses of all peaks occurring in S ₁ and S ₂ .

ξξ _{i, j} (i = 1..n, j = 1: 2) is the matrix of intensities.

The intensity to earth i for the centroid gets one accordingly as the average of the intensities in columns 1 to 2.

For superspectrums and centroids of isolates that are not identical to one of the reference strains or to another isolate, it is checked whether their distance to all available super spectra and centroids is greater than σ ₂ . If this is the case, the isolate is detected as new subspecies.

If the distance of the superspectral spectrum is a single culture to a superspec spectrum of another single culture or of a reference subspecies between σ ₁ and σ ₂ , the single culture is considered to be "closely related, possibly identical" to the other single culture or reference subspecies.

The Distance matrix allows on the one hand a pairwise comparison of the Spectra of reference subspecies or single cultures with indication a quantitative similarity measure and on the other hand forms the basis for the application of different clustering methods, prefers hierarchical clustering, K-means clustering, fuzzy clustering and quantum clustering (Hoppner et al., 1999; Hastie et al., 2003, Horn & Axel, 2003). These procedures serve the similarity relationships of all investigated subspecies to capture each other simultaneously.

Preferably, the superspeciences or centroids of the individual cultures remaining after the pairwise comparison are now characterized together with those of the reference subspecies by means of different methods of cluster analysis with regard to their phenotypic similarities. By means of hierarchical Clustering, K-means clustering or quantum clustering, the super-spectrum of each individual culture or centroid is clearly assigned to a cluster.

For example, the distance matrix is analyzed by agglomerative hierarchical clustering using the farthest neighbor method, which is based on the maximum distance between the superspectrons (Hastie et al., 2003). To assess the quality of clustering, Cophenetic correlation coefficients are calculated between the individual clusters. The results are considered correct if the distances between the super spectra within the clusters correlate with the distances calculated in the distance matrix d _mn . Low values of the cophenetic correlation coefficient are excluded as artifacts. The results are displayed in dendrograms.

At the Quantum clustering e.g. becomes the matrix in a first step converted into eigenvectors and eigenvalues. Thereafter, preferably the "singular Value Decomposition "method applied. For the reduction of the matrix only becomes the determining eigenvalues used. Through this method, the superspectrons can pass through Vectors in a space of lesser dimensionality are replaced. The parameter σ which characterizes the relative cluster size, is adjusted on an empirical basis. The method determines a variable number of clusters and assigns each superspectrum to exactly to one of them. The reliability of the method is through Cross-validation by means of "jack-knife testing "(omitting of a value).

Additionally performed fuzzy Clustering allowed for each individual culture the probability of correct assignment to investigate. In all cases serves stability the connections between the reference subspecies as an internal control. The distribution the superspectrons of the individual cultures on different clusters, whose Distance and inner structure form the basis of differentiation Subspecies both in terms of individual properties each individual culture as well as the population structure of all individual cultures from a biological sample.

Prefers becomes the method according to the invention for differentiation to subspecies subsidence in pathogenic bacteria, For example, oral bacteria such as streptococci, applied. Especially preferred is the use of the method of differentiation of mutans streptococci, in particular for the differentiation of subspecies of the species Streptococcus mutans or Streptococcus sobrinus.

The Method is also suitable for the differentiation of unicellular Mushrooms, like oral mushrooms. For example, the method can be used for Differentiation of subspecies of Candida albicans used. It can be used as a selective medium "Chromagar Candida "(Mast Diagnostica GmbH, Reinfeld).

Based an embodiment the invention will be closer explained.

Embodiment 1

Differentiation from subspice isolates of Streptococcus recovered from human saliva mutans

to Differentiation of subspecies isolates of Streptococcus mutans from human saliva following steps are performed:

1. Cultivation of the reference subspecies of Streptococcus mutans

The reference strains Ingbritt, JB 1600, NCTC 10449, OMZ 125, GS 5, SE 11, OMZ 175, LM 7, QP 50-1 from S. mutans are on Mitis-Salivarius agar with Bacitrazin (MSB, Difco) a selective medium for human pathogenic mutans streptococci, at 37 ° C cultured.

2. Cultivation of in the biological sample contained microorganisms

each The 10 test persons chew about 1 minute on a commercial Paraffin block for Saliva tests and collects the resulting saliva. A lot 0.1-1 ml of saliva is added to Mitis-Salivarius Agar with Bacitrazin, a selective medium for human pathogenic mutans streptococci, streaked and incubated at 37 ° C for 24-48 Incubated for hours.

The Isolates are determined by means of biochemical tests (colorful series) and above it Membrane fatty acid spectrum examined. Isolates belonging to the species S. mutans become mass spectrometric examined.

3. Sample preparation for MALDI-TOF-MS

Separate Cell colonies of S. mutans reference strains and isolates from oral cavity are carefully washed one to two times in purified water. The colonies are freshly prepared MALDI-matrix solution (α-cyano-4-hydroxy-cinnamic acid, (HCCA) in acetonitrile / 0.1% trifluoroacetic acid (TFA) (2: 1)). 1 μl of this Suspension is placed on a MALDI sample carrier and dried. The sample is filled with 5 μl 0.1% TFA washed to salt and any residues of the Culture medium dissolve. The Sample is recrystallized using acetonitrile / 0.1% TFA (2: 1).

alternative can The cells are also placed directly on a slide and on this with MALDI matrix solution be offset. The further procedure is carried out as already described.

It several samples are applied from one and the same colony (several spots).

4. Recording the MALDI-TOF mass spectra

The MALDI sample carrier is introduced into a MALDI-TOF mass spectrometer (Bruker Biflex III, N ₂ laser (wavelength 337 nm, pulse 5 ns, flight distance 1200 mm).) The measurements are in linear positive mode (deceleration 400 ns, voltage 20 kV , Mass range: 2000 to 20,000 daltons) For each sample spot, a total of 120 shots (4 picking cycles of 30 shots each) are applied.

For calibration, a peptide mix (Sigma, Taufkirchen) is used (somatostatin 28 (synthetic, M ⁺ H ⁺ = 3148.0 daltons), insulin (bovine pancreas, M ⁺ H ⁺ = 5733.5 daltons), cytochrome C (horse heart, M ⁺ H ⁺ = 12360.0 daltons) and myoglobin (equine heart, M ⁺ H ⁺ 16951.0 daltons) The spectra are calibrated internally.

It become multiple sum spectra of different spots of the same Sample taken. Six of the sum spectra are superimposed, higher spectra To obtain reproducibility.

From 120 mass / charge peaks are detected in these superspectrons. lists which contain the mass / charge peaks and their intensities are called ASCII Files into the MatLab 6.5 program (The MathWorks Inc., Natick, U.S.A.). All further steps are under Using the corresponding MATLAB functions computer-aided execution.

1 shows super spectra of S. mutans reference strains.

5. Classification and Identification of peaks in super spectra

The Peak intensities every superspeed is log-transformed by means of a "Loess smoothers " set the baseline.

To compare different super spectra, a mass window is guided along the x-axis (m / z). The center of this mass window represents the mean position of the comparable peaks in all spectra studied. The window width is determined by the precision of the experimental m / z determination. A linear relationship to m / z is assumed. Thus results: Window width = absolute width + relative width · peak mass.

For the window width 3 Da, for the relative width 2 e ^{-4 were} determined. The mass window is assigned to each mass peak. Windows containing the same peaks in different super spectra are combined. These peaks are represented by their mean m / z and their mean intensity. Overlapping windows are resolved with half the window width.

The described method constructs a matrix η _ij . Their lines are determined by the middle peak masses, their columns by the respective superspec spectrum. In the absence of a peak in a superspec spectrum, η _ij = 0 was assumed. A typical matrix contains 100-200 lines. The number of columns is equal to the number of superspectrons analyzed.

6. Create the distance matrix and analysis of the distance matrix using various clustering methods

The similarity of the superspec spectra of the reference subspecies is represented by a Euclidean distance matrix. The distance matrix is analyzed using different clustering methods (Hierarchical Clustering and Qantum Clustering). The results for the clustering of the superspectrons of the reference strains are in 2 shown. Both clustering methods allow a secure and identical separation of the reference strains.

From each reference subspecies, two independent super spectra were determined: 1, 2: Ingbritt 3, 4: JB 1600 5, 6: NCTC 10449 7, 8: Gs 5 9, 10: QP 50-1 11, 12: SE 11 13, 14: OMZ 125 15, 16: LM 7 17, 18: OMZ 175

2a shows the result after hierarchical clustering.

lower horizontal line: σ ₁ , maximum distance of the double determination of a subspecies upper horizontal line: σ ₂ , minimum distance of two different subspecies

2 B shows the result after quantum clustering, where the number of dimensions is 9.

The Subspecies of the species S. mutans show different similarities, however, they are lower (higher Differences) than the largest differences, observed between duplicate determinations of a strain.

7. Subspecies differentiation of isolates

The superspecific analysis of S. mutans isolates at subspecies level is based on the reproducibility of measurements on the S. mutans reference strains. 2a shows the results of the hierarchical clustering of the superspectrons of the reference strains, which were each determined twice. σ ₁ is a measure of the maximum difference of the double determination of a subspecies (here the double determination of the strain SE 11). σ ₂ is a measure of the minimum distance between superspectrics of different subspecies (here the distance of the strains Ingbritt and JB 1600).

In a first step, all S. mutans isolates of each subject were analyzed separately. Using σ ₁ as a barrier, the maximum number of different subspecies per subject could be determined. With σ ₂ as a barrier between 1 and 4 different subspecies could be determined for each subject (see Table 1).

Table 1: Result of the population analysis of clinical S. mutans isolates of all 10 subjects at subspecies level.

3 shows the results of the hierarchical clustering of the superspec- ces or centroids of all surely different S. mutans subspecies of each of the 10 subjects, the double determined superspec- ce of the reference strain SE 11 (maximum difference of the double determination of the reference strains, σ ₁ ) as well as the super spectra of the two most similar different reference strains (Inbritt and JB 1600, σ ₂ ).

With σ ₁ as a barrier isolates of different subjects could be identified as phenotypically indistinguishable, and thus within the scope of the method used as "phenotypically identical" (here are subspecies a from subject 3 (P 3-a) and subspecies a from subject 6 (P 6-a) identical as well as subspecies a from subject 6 (P 6-a) and subspecies a from subject 5 (P 5-a)).

With σ ₂ as a barrier it was possible to distinguish at least 14 distinct subspecies in the 10 subjects.

Distances between σ ₁ and σ ₂ were found for three groups of isolates (P 3-a / P 6-a, P 1-a / P 5-a, P 9-a, P 4-a, P 6-b; P 1-b, P 8-a, P 7-a, P 2-a, P 3-c; 3 ). The respective phenotypes are distinguishable, but the respective super spectra are more similar to each other than those of the most similar of the reference strains studied (Ingbritt and JB 1600).

Literature:

• Allmaier G, Schaffer C, Messner P, Rapp U, Mayer-Posner FJ. Accurate determination of the molecular weight of the major surface layer protein isolated from clostridium thermosaccharolyticum by time-of-flight mass spectrometry. 1995: J Bacteriol 177: 1402 to 1404.
• Bright JJ, Claydon MA, Soufian M, Gordon DB. Rapid typing of bacteria using matrix-assisted laser desorption ionization time-of-flight mass spectrometry and pattern recognition software. J Microbiol Methods 2002: 48: 127-138.
• Elhanany E, Barak R, Fisher M, Kobiler D, Altboum Z. Detection of Bacillus anthracis spore biomarkers by matrix-assisted laser desorption / ionization time-of-flight mass spectrometry. Rapid Commun Mass Spectrom. 2001; 15 (22): 2110-6.
• T. Hastie, R. Tibshirani, J. Friedman (2003) The Elements of Statistical Learning: Data Mining, Inference, and Prediction Series. Springer Series in Statistics; Springer-Verlag.
• Hathout Y, Setlow B, Cabrera-Martinez RM, Fenselau C, Setlow P. Small, acid-soluble proteins as biomarkers in mass spectrometry analysis of Bacillus spores. Appl Environ Microbiol. 2003 Feb; 69 (2): 1100-7.
• F. Hoppner, F. Klawonn, R. Kruse and Th. Runkler. fuzzy Cluster Analysis, Methods for Classification, Data Analysis and Image Recognition. John Wiley & Sons Ltd, 1999.
• D. Horn, I. Axel (2003) Bioinformatics, 12, 1110-5. Novel Clustering algorithm for microarray expression data in a truncated SVD space.
• Hathout Y, Demirev PA, Ho YP, Bundy JL, Ryzhov V, Sapp L, Stutler J, Jackman J, Fenselau C. Identification of Bacillus spores by matrix-assisted laser desorption ionization-mass spectrometry. Appl Environ Microbiol. 1999 Oct; 65 (10): 4313-9.
• Kallow, W., Dieckmann, R., Kleinkauf, N., Erhard, M., and Neuhof, T. (2002), Offenlegungsschrift, German Patent and Trademark Office Munich, Az .: 100 38 694.6.
• Kato M. Characterization of fatty acid composition in the cytoplasmic membrane of Streptococcus mutans. Gifu Shika Gakkai Zasshi. 1989 Jun; 16 (1): 16-39.
• McNall RJ, Adang MJ. Identification of novel Bacillus thuringiensis Mandluka Sexta midgut through proteomic CrylAc binding proteins analysis. Insect Biochem Mol Biol. 2003 Oct; 33 (10): 999-1010.
• Stößer L, Kneist S, Heinrich-Weltzien R, Fischer T, Tietze W Current Research on Caries Risk Assessment. In: Stookey, GK (ed.): Early Detection of Dental Caries II: Proceedings of the 4 ^th Annual Indiana Conference Indianapolis, India. Indianapolis, IN: Indiana University School of Dentistry, 2000, pp. 31-56.
• Whiteaker JR, Warscheid B, Pribil P, Hathout Y, Fenselau C. Complete sequences of small acid-soluble proteins from Bacillus globigii. J Mass Spectrom. 2004 Oct; 39 (10): 1113-21.

Claims (8)

Method for the differentiation of microorganisms in a sample at sub-species level by means of matrix-assisted laser desorption / ionization time-of-flight mass spectrometry (MALDI-TOF MS), a) being obtained from the sample by culturing on species-selective medium individual cultures of microorganisms, the belong to the same species, b) MALDI-MS spectra of the individual cultures obtained from the sample are recorded, c) by a mathematical similarity analysis from MALDI-MS reference spectra of different microorganism subspecies that belong to the same species as the individual cultures obtained from the sample d) the values σ ₁ (homogeneity limit) and σ ₂ (heterogeneity limit) are determined, d) the MALDI-MS spectra of the individual cultures obtained from the sample are compared in pairs with each other and with the reference spectra by means of a mathematical similarity analysis, the individual culture being identical to a reference subspecies or another from de If the distance of the spectra is smaller than σ ₁ , the individual culture obtained is a new subspecies, if its distance to all reference spectra and to all spectra of the other single cultures obtained from the sample is greater than σ ₂ , which is close to single culture is related to a reference subspecies or other single culture obtained from the sample if the distance of the spectra is greater than σ ₁ and less than σ ₂ Method according to claim 1, characterized in that that the MALDI-TOF-MS spectra in a mass range of 2000 to 20000 daltons are recorded. Method according to claim 1, characterized in that that the MALDI-TOF-MS spectra from the superposition of at least six complete Single spectra are constructed. A method according to claim 1, characterized in that for each reference subspecies at least two independent spectra are recorded. A method according to claim 1, characterized in that the values σ ₁ and σ _{2 are obtained} from a Euclidean distance matrix of the reference spectra. Method according to claim 1, characterized in that that the MALDI-MS spectra of the individual cultures obtained from the sample by means of a Euclidean distance matrix in pairs with each other and be compared with the reference spectra. Method according to claim 1, characterized in that that in addition a cluster analysis of the MALDI-MS spectra obtained from the sample Single cultures and / or the reference spectra is performed. Method according to claim 7, characterized in that that for cluster analysis hierarchical clustering, K-means clustering, Fuzzy clustering and / or quantum clustering is used.